Has MySQL Forked Beyond Repair?

snydeq writes "Fatal Exception's Neil McAllister questions the effect recent developments in the MySQL community will have on MySQL's future in the wake of Oracle's acquisition of Sun. Even before Oracle announced its buyout, there were signs of strain within the MySQL community, with key MySQL employees exiting and forks of the MySQL codebase arising, including Widenius' MariaDB. Now Widenius' Oracle-less Open Database Alliance adds further doubt as to which branch of MySQL will be considered 'official' going forward. 'Forks are a fact of life in the open source community, and arguably an entirely healthy one,' McAllister writes. 'Oracle just better hope it doesn't end up on the wrong side of the fork.' To do so, he suggests Oracle will have to regain the the trust and support of the MySQL community — in other words, 'stop acting like Oracle.'"

Comments

Cue postgres fan bois

[Anonymous Coward]

In 3, 2, 1

Re:Cue postgres fan bois

[danomac]

It's not so much fanboys. When you develop for something, you want to develop for a stable feature set. If there's going to be a dozen forks of a database, it becomes much more work to test all the versions and apply patches.

It now makes far more sense to develop using a different DBMS.

Note: I've never installed or used PostgreSQL.

Re:Cue postgres fan bois

[Anonymous Coward]

Both MySQL and PostgreSQL are junk compared to Microsoft SQL Server.

Re:Cue postgres fan bois

[Moblaster]

It is more accurate to characterize the recent burst of minor MySQL variations as a sporking of the code base.

Re:Cue postgres fan bois

[mikael_j]

Clearly this is a troll but regardless of all other features that MSSQL may have that MySQL doesn't there is one thing that is missing: LIMIT. No, TOP is useless for a lot of stuff and the fact that LIMIT is missing results in lots of painful hacks with nested SELECTs and other crap in stored procedures.

/Mikael

Re:Cue postgres fan bois

[amicusNYCL]

That's entirely correct. MSSQL has several nice features, but the lack of LIMIT.. well.. really limits its usefulness. The concept of a LIMIT clause is so useful that the lack of it is essentially a dealbreaker. The hacks to implement some sort of limiting using TOP are very inefficient and don't result in any sort of performance increase comparable to the one you get with LIMIT.

Re:Cue postgres fan bois

[iluvcapra]

When you develop for something, you want to develop for a stable feature set.

Or in the case of MySQL, a featureset at all. I keed! I keed!

Re:Cue postgres fan bois

[$RANDOMLUSER]

Both MySQL and PostgreSQL are junk compared to Microsoft SQL Server.

No, that's YourSQL.

Re:Cue postgres fan bois

[houstonbofh]

Both MySQL and PostgreSQL are junk compared to Microsoft SQL Server.

No, that's YourSQL.

No, it's Bill's SQL.

Re:Cue postgres fan bois

[K. S. Kyosuke]

Come to think of it, MySQL had been forked on the inside long before that - a thousand and one storage engine and none of them complete... (Just stating the facts.)

Re:Cue postgres fan bois

[Jurily]

When you develop for something, you want to develop for a stable feature set. If there's going to be a dozen forks of a database, it becomes much more work to test all the versions and apply patches.

The very existence of these forks proves MySQL could easily be better, and/or there are major problems with the development method. Either way, it's a sinking ship. OTOH, a dead project provides the ultimate stability.

Re:Cue postgres fan bois

[larry bagina]

Why doesn't SQL server have BEFORE triggers? Or row-level triggers?

Re:Cue postgres fan bois

[Anonymous Coward]

Dude, maybe you should wake up: SQL Server supports ANSI SQL windowing function ROW_NUMBER() OVER (ORDER BY) and, of course, when you apply a predicate to the resulting row number all optimization are done.

Then again if LIMIT is The Thing that make you choose a DBMS you may as well get a b-tree indexing library and play around with offset by yourself...

Re:Cue postgres fan bois

[Locke2005]

One thing I can say for SQL Server: it sure beats the hell out of Access!

Re:Cue postgres fan bois

[gazuga]

I only come across it occasionally, but several times I have found myself wishing for LEAST and GREATEST functions in T-SQL, like in Oracle. Of course people always suggest creating workarounds via a CLR addin, functions or CASE statements. But really, in both cases, this seems like a trivial feature to implement. C'mon MS!

Re:Cue postgres fan bois

[Rhapsody Scarlet]

Both MySQL and PostgreSQL are junk compared to Microsoft SQL Server.

No, that's YourSQL.

No, it's Bill's SQL.

Isn't it really more Steve's SQL these days?

[Insert overdone chair-throwing joke here]

Re:Cue postgres fan bois

[Anonymous Coward]

If LIMIT is The Thing, why would you assume he knows what is a b-tree indexing library?

Re:Cue postgres fan bois

[Anonymous Coward]

Why is this modded troll? It's been shown to be true in benchmark after benchmark.

Re:Cue postgres fan bois

[shutdown -p now]

Clearly this is a troll but regardless of all other features that MSSQL may have that MySQL doesn't there is one thing that is missing: LIMIT.

Perhaps it's because LIMIT isn't ANSI SQL? Before SQL-2008, you were supposed to use ROW_NUMBER() for that, which MSSQL does support. And no-one supports SQL-2008 yet.

Re:Cue postgres fan bois

[Phizzle]

Both MySQL and PostgreSQL are junk compared to Microsoft SQL Server.

No, that's YourSQL.

No, it's Bill's SQL.

Wait, isnt MySQL now Larry's SQL?!
*teh zing!*

Re:Cue postgres fan bois

[Anonymous Coward]

Even with a CLR function you are out luck because GREATEST can take an arbitrary number of parameters and this is something you can't do in SQL Server. Even if you define a fixed number of parameters, you still have to specify the "default" keyword when calling the function which is not exactly practical. Moreover, you'll have problem with data types. So yes, there are still nasty limitations compared to Oracle, but to be honest I never found that much need for these 2 specific functions.

Re:Cue postgres fan bois

[itlurksbeneath]

Considering that they just licensed the technology from Sybase, isn't that "We couldn't write it ourselves (look how Access turned out) so we bought it"SQL?

Re:Cue postgres fan bois

[drinkypoo]

It's Bill's SQL which is based back in the way-back on Sybase 10 and is one of the few apps Microsoft hasn't totally trashed since acquisition. The management tools are painful, but no moreso than say Oracle. (Oracle does a lot more, but the pain, oh the pain.)

Re:Cue postgres fan bois

[Anonymous Coward]

Assuming you have a unique key on the table called ID,

SELECT * FROM Table LIMIT 20, 10

becomes

SELECT TOP (10) * FROM Table WHERE ID > 20 ORDER BY ID

?

Re:Cue postgres fan bois

[Cico71]

Why doesn't SQL server have BEFORE triggers?

Well, you can use INSTEAD OF triggers. But you still have to do the actual operation in the trigger body, so yes it would be nice if they implemented it. Reality is that features gets prioritized and although it may seem a very popular thing, it's actually not.

Or row-level triggers?

Because the preferred approach with (almost) relational technology is set based instead of row-based? You can still have a row based approach in the body of the trigger if you want with a cursor and a loop. However, in most cases you simply deal with the whole set (inserted and deleted virtual tables instead of :new and :old virtual aliases). Maybe I should ask why in Oracle the examples for statement based triggers are always the same: they log the fact that an action triggered, they never deal with the data...

Anyway, a couple of better trigger questions (or rather a better flaming questions) are:

• Why MSSQL doesn't support trigger with predicates? (where conditions)
• Why it doesn't support columns triggers?

In both cases you have to test conditions inside the trigger body, but the trigger always fires.

Want other, better, flaming questions?

• Sequences?
• Autonomous transactions?
• Packages?
• Polymorphism for functions and procedure?

and so on....

Re:Cue postgres fan bois

[ppanon]

OTOH, a dead project provides the ultimate stability

Not when it comes to security vulnerabilities.

Re:Cue postgres fan bois

[Anonymous Coward]

Assuming you have a unique key on the table called ID,

SELECT * FROM Table LIMIT 20, 10

becomes

SELECT TOP (10) * FROM Table WHERE ID > 20 ORDER BY ID

?

No, this is not the same. Your example relies on the fact that you have a monotonically increasing numeric attribute called ID without gaps in the numbering.

It may be the case, or it may be not. To the point that, afaik, you can LIMIT just relying on physical ordering (sic).

The correct implementation would be something like

SELECT * FROM (SELECT ROW_NUMBER() OVER (ORDER BY id) as rn, * FROM dbo.test_deadlock ) AS T WHERE T.rn BETWEEN 20 and 20+10

Of course you always have to provide and attribute for the ORDER BY otherwise it doesn't really make any sense to "limit" the result set in the first place (unless someone is so dumb to rely on physical ordering defeating completely the idea of having a DBMS with an optimizer etc. etc.)

Re:Cue postgres fan bois

[Anonymous Coward]

Have you USED SQL Server? Have you USED Enterprise Manager? Do not say "Yes" Because making the comment you did proves you have not, or that you are an imbecile. Enterprise Manager is quite possibly the single best thing Microsoft has ever turned out. Were it able to interface with other Databases with a fraction of the functionality it currently has, MS would be selling a copy to every damned database developer who values their time. SQL Server might not be ORACLE for scalability, but it comes real close, and it sure is a hell of a lot more scalable than MySQL, and yes, I am/was a MS Certified DBA who now develops apps based on MySQL. I am not fanboi, but I DO know both. SQL Server FTW without doubt

Re:Cue postgres fan bois

[Anonymous Coward]

MSSQL supports the (cumbersome) ISO rownumber operator doesn't it? The syntax to apply a limit is a bit bodgy, but it does work neh?

This gets rows 100 through 200.

SELECT * FROM (
                              SELECT
                                          ROW_NUMBER() OVER (ORDER BY key ASC) AS rownumber,
                                          columns
                                    FROM tablename
                          ) AS foo
          WHERE rownumber > 100 and rownumber = 200;

Re:Cue postgres fan bois

[Anonymous Coward]

It's Bill's SQL which is based back in the way-back on Sybase 10 and is one of the few apps Microsoft hasn't totally trashed since acquisition.

No, you can read up the whole development story in "Inside SQL Server book" if you want. Sybase code base was used up to version 6.5 of SQL Server. Starting from version 7.0 the product was completely re-engineered to the point that, AFAIK, there's nothing left from the original code base in today's versions. But maybe you can surprise me with some references...

The management tools are painful, but no moreso than say Oracle. (Oracle does a lot more, but the pain, oh the pain.)

Come on, this is simply wild trolling. So you got an "Insightful" for false statements and trolling, I guess this is how it works with some guys around here...

Re:Cue postgres fan bois

[uassholes]

Is there a reference for the differences between MS's Sybase and Sybase's Sybase these days?

Re:Cue postgres fan bois

[AC-x]

There is a work around using ROW_NUMBER in MSSQL that's just as fast but the sql you need to write is U-G-L-Y

Re:Cue postgres fan bois

[Anonymous Coward]

Can I see MySQL implementation of recursive/hierarchical queries and windowing functions?

I'll trade analytical functions over syntactic sugar to page record sets anytime.

Re:Cue postgres fan bois

[afidel]

Actually MS wrote one of the most stable and reliable database engines ever, JET Blue aka ESE. It's the multimaster replication database behind Active Directory and there are almost never problems with it. It also backs Exchange and when they worked out the application code problems (in Exchange 2007) it's really shined there as well.

Oh yeah and supposedly SQL Server 2005 was a ground up rewrite, I doubt there's much code left over from pre-SQL 7.

Re:Cue postgres fan bois

[timmarhy]

if i'm a fan boi because i prefer the better database, then guilty as charged sir...

Re:Cue postgres fan bois

[Evets]

I've used it plenty. I've scaled MySQL better, and more importantly - without any question as to how much I need to pay for licensing.

Enterprise Manager has changed remarkably little since it was taken over by Microsoft.

DTS portability and authoring is poor within EM to say the least. SQL Agent scheduling of DTS scripts is lacking as well. Querying, user management, etc. are all more intuitive than Oracle, but still the interface is quite poor.

MS SQL has it's place. But it certainly doesn't separate itself as significantly better than OSS solutions for 90% of database usage scenarios (at least 90% of the ones that I've been involved with).

Re:Cue postgres fan bois

[AmberBlackCat]

If that's true then all you have to do is ignore the forks...

Re:Cue postgres fan bois

[Requiem18th]

I had to work with MS SQL for almost two years for a living and used enterprise manager daily and I have to say that I was terribly unimpressed. Granted, I didn't use all of it's features, but it largely struck me as a clutch around the incompetence of its users.

The most impressive uses I saw for it was using it to pipe data to another server and scheduling backups. In Linux I could do both with cron/python or simple mysql scripts or a web service and apache etc.

Enterprise manager is essentially another "programming by gui" tool for people who can't use a command line or write scripts who surprisingly --considering they are programmers-- constitute the majority of my coworkers, except for a java guy (a solaris fan) and a php guy (another linux fan).

Needless to say we used to run circles around the other developers, and I am a lazy ass slacker.

Re:Cue postgres fan bois

[Anonymous Coward]

So does snorting cockroaches.

Re:Cue postgres fan bois

[Blakey Rat]

Enterprise Manager has changed remarkably little since it was taken over by Microsoft.

Hasn't Enterprise Manager been completely obsolete for the last two releases? You're supposed to be using SQL Server Management Studio. (I don't know if you'd consider that better or worse, but... at least it's "remarkably changed" from Enterprise Manager.)

Re:Cue postgres fan bois

[Anonymous Coward]

At my job we needed GREATEST and LEAST for datetime types, so we just wrote that as an user-defined function in T-SQL; no need to use CLR. The two-argument limitation is only a minor annoyance; we just nest calls to the functions to get around it.

But overall, yeah, Oracle's set of built-in functions is often nicer than SQL Server's and DB2's.

Re:Cue postgres fan bois

[kilodelta]

Say what? In my experience an MS-SQL server could NOT keep up with heavy hits for a corporate database. We had to replicate the MS-SQL tables to MySQL in order not to have to restart a server every 30 minutes.

Re:Cue postgres fan bois

[ScrewMaster]

Both MySQL and PostgreSQL are junk compared to Microsoft SQL Server.

No, that's YourSQL.

No, it's Bill's SQL.

Otherwise known as the "Structured Queerly Language."

Re:Cue postgres fan bois

[iluvcapra]

From +4 funny to +2 flamebait in 2 hours... Touchy crew here. Maybe I hit a little too close to the mark.

Re:Cue postgres fan bois

[BPPG]

if it makes you feel any better, I laughed.

Re:Cue postgres fan bois

[darthvader100]

And non can compare to the Oracle (Oracle Fanboi alert) ;)

Re:Cue postgres fan bois

[redhog]

Oracle causes servere brain pain. Just a fun example:

In Oracle, NULL and empty string '' are the same value. But (NULL == NULL) IS NULL in ANSI SQL. So in Oracle, ('' == '') IS NULL, not TRUE...

If you don't think this is a problem, imagine this:

create table person (name varchar(32), address varchar(32));
insert into person (name, address) values ('Johan', '');
insert into person (name, address) values ('Elsa', 'SWEDEN');
insert into person (name, address) values ('Sanna', '');

And then, in some GUI app, someone has the code:

cursor.execute("select * from person where address = ?", addr_form_gui)

Now, the user using this apps tries to find problematic entries - persons w/o address, and enters an empty string for address. And gets no resluts at all... WOW.

Re:Cue postgres fan bois

[MariusBoo]

I work with MSSQL daily now and I am impressed with Management Studio (the app that replaced query browser and enterprise manager). OK, maybe impressed is not the right word but it's still much better than PgAdmin or MySQL Admin (or what it's called).

It also has a lot of really nice features: out of the box replication, log shipping, mirroring (which rocks). The only things that I would really like to have from Postgres are query rewriting and sequences(I would really, really like sequences instead of this identity shit).

Anyway if it were free I would definitely choose it every time over Postgres or MySQL.

Re:Cue postgres fan bois

[rbrausse]

hey, Access is a nice GUI for MSSQL :)

Re:Cue postgres fan bois

[richlv]

ChairSQL.
if you do things it doesn't like, it hurls a table at you, though.

Re:Cue postgres fan bois

[maxwell demon]

OTOH, a dead project provides the ultimate stability

Not when it comes to security vulnerabilities.

OTOH, a dead project provides the ultimate stability

Not when it comes to security vulnerabilities.

It's stable in respect to security vulnerabilities as well. No need to create new exploits just to keep your malware compatible.

Re:Cue postgres fan bois

[remmelt]

Can you write an example SQL statement that does the same as

SELECT * FROM customer ORDER BY last_name LIMIT 5, 10

This shows records 6-10 from the customers table.

Re:Cue postgres fan bois

[Anonymous Coward]

Already posted here: http://developers.slashdot.org/comments.pl?sid=1241297&cid=28047331 change id with last_name and range values as needed.

Re:Cue postgres fan bois

[AftanGustur]

It's not so much fanboys. When you develop for something, you want to develop for a stable feature set. If there's going to be a dozen forks of a database, it becomes much more work to test all the versions and apply patches. It now makes far more sense to develop using a different DBMS.

Microsoft has said exactly the same thing about Linux ..

What will happen is that developers will support one "brand" of MySQL. Just like things are today with Linux distributions.
Most corporate offerings support Only Redhat Enterprise Linux,or SuSE, although their products work (and are run on) different distributions.

That will also probably be the case with MySQL.

Re:Cue postgres fan bois

[Cico71]

Interesting... but it seems NASDAQ and the London Stock Exchange, among several others, do not agree with your field experience.

As another example take the Bwin case study:

The bwin Data Management Systems group uses SQL Server 2008 to continue its tradition of providing world-class performance for its sports betting customers. During peak loads SQL Server handles more than 6,000 financial transactions per second, which Grohser says translates into more than 30,000 database transactions per second.

Re:Cue postgres fan bois

[MariusBoo]

What the other guy in the same thread says. You are supposed to author Integration Services Packages (the new DTS) using Business Intelligence Studio. Scheduling the packages is pretty straightforward and it includes options for using windows authentication different than the SQL context (run under a different windows user).

Integration Services (along with Analysis Services) is one of the best done things to come out of Microsoft. It's easy, reasonably powerful and integrates completely with the rest of SQL Server suite.

Re:Cue postgres fan bois

[CrashandDie]

SELECT * FROM customer ORDER BY last_name LIMIT 5, 10

This shows records 6-10 from the customers table.

No it doesn't. It's going to show records 6 - 15.

Flame all you want, but at least do it properly.

Re:Cue postgres fan bois

[Anne Thwacks]

Both MySQL and PostgreSQL are junk compared to Microsoft SQL Server.

Have they finally fixed date handling in MSSql server? Personally, I'd say MySQL and MSSql are junk compared to PostgreSQL, but that is just because I have used all three. (I have also used DB2 which beats all three, and Oracle, which is just annoying).

Re:Cue postgres fan bois

[bytesex]

No that doesn't fix it. That requires you to know the range you're trying to retrieve. When you don't know that (there might be 10 results from 1 to 10, but 700000 results from 10 to 20), the fact that the ROW_NUMBER() function is part of the WHERE clause is the dealbreaker: matching rows get fetched first (in arbitrary order) and only then sorted, not, as you would want, sorted first and then only have a subset delivered.

Re:Cue postgres fan bois

[Cico71]

I'm sorry but I don't understand you. If the results are 10, they are marked from 1 to 10. If the results are 700000, they are marked from 1 to 700000. I don't know where you take the 10 to 20, a row number is just a row number and has no relationship with other ids or numbers (except for possible ordering, but see next paragraph).

Moreover, I can't make any sense of what you're saying about matching rows fetched in arbitrary order and so on. Look is very simple: row numbers are logically assigned in the inside table expression (query if you wish) based on the specified ordering in the OVER (ORDER BY) clause (and not in arbitrary order). Then rows are filtered in the where clause based on the give row number (so order is completely preserved). Please note that I said "logically" because, of course, the whole result of the table expression is not completely materialized before applying the outer where clause. Streaming operators in the query plan permit access optimizations so that the query process only the minimum amount of data.

Re:Cue postgres fan bois

[cheekyboy]

nice for you, but unless you document every script and how to use it, no one knows how you run your shop or can take over.

Sure you could use a fancy gui trigger with tk to call your scripts, or even java.

Nothing wrong with guis either, scripts are nice, but arent the only way to automate things.

Re:Cue postgres fan bois

[asdf7890]

You're supposed to be using SQL Server Management Studio. (I don't know if you'd consider that better or worse, but... at least it's "remarkably changed" from Enterprise Manager.)

I wouldn't agree there. From what I see SSMS is basically the old Enterprise Manager and Query Analyzer programs integrated, with some useful but minor UI improvements along the way. Oh, and a simple "project" abstraction for grouping script files and connection settings, and hooks to make other tools look more integrated.

Not what I would call remarkably changed. Certainly better, but an evolution rather than a revolution.

Caveat: I'm using SSMS as provided with SQL2005, we have yet to move towards SQL2008 so for all I know there could be more siginicant changes in the latest revision of the tool.

Re:Cue postgres fan bois

[kilodelta]

I do not trust a study done by Microsoft. I'd much prefer an independent agency. Granted, their newer product (MS-SQL 2008) might have solved some of the issues had with older.

My fundamental issue with MS and databases is that it's just too expensive in terms of money and hardware.

Re:Cue postgres fan bois

[Thundersnatch]

In my experience an MS-SQL server could NOT keep up with heavy hits for a corporate database. We had to replicate the MS-SQL tables to MySQL in order not to have to restart a server every 30 minutes.

I'd say you really don't know what you're doing. SQL 7.0 had some odd locking behavior, but that is over 10 years old. We have one SQL 2005 application that supports hundreds of thousands of users and 1M+ transactions per day. On a single 8-core box with 16 GB of RAM (with warm standby in case of hardware failure). SQL gets restarted about once every six months for a security patch. And it requires almost no tuning, at least compared to Oracle.

SQL Server is clearly Microsoft's best product, and probably the best general SQL database available in terms of feature/performance/usability balance.

Re:Cue postgres fan bois

[geminidomino]

Enterprise Manager is quite possibly the single best thing Microsoft has ever turned out

"Turned out?" Didn't they borg it from someone else?

Re:Cue postgres fan bois

[Anonymous Coward]

OTOH, a dead project provides the ultimate stability

Not when it comes to security vulnerabilities.

Actually, when it's dead the security vulnerabilities are as stable as can be as well ;-) They'll always be there!

Re:Cue postgres fan bois

[kilodelta]

Nah, it was MS-SQL 7. State government is notorious for using old software.

I agree on Oracle though - it's why I embrace MySQL and am very concerned by the current turbulence in ownership of the product.

Re:Cue postgres fan bois

[artemis67]

It's completely forked.

But now, who's forked more, the software or the users?

Well, fork me, I don't have a forking clue.

Re:Cue postgres fan bois

[Anonymous Coward]

This is incorrect. At one point Microsoft rewrote the whole product. I believe that was SQL Server 7.0.

Re:Cue postgres fan bois

[jadavis]

a thousand and one storage engine

Those aren't exactly forks, but you bring up an interesting point: the different storage engines don't support the same features, and don't have the same user-visible behaviors. So, that means application developers must test on all of the various storage engines that they intend to support -- which is very similar to supporting many forks of the same product.

Re:Cue postgres fan bois

[Blakey Rat]

The 2008 version is further evolved, but nothing completely breathtaking.

What kind of changes are you expecting to see? You're right that, fundamentally, SSMS is just Enterprise Manager and Query Analyzer combined, but it's also a pretty damned mature field. What kind of revolution are you waiting for?

Re:Cue postgres fan bois

[Requiem18th]

Are you implying that GUI task...icon thingies don't need to be documented? Please, I had to document them as well or nobody else would even know they were there.

Yes scripts aren't the only way to automate things but we are programmers! they are the natural way for us! And it is the most flexible way too, Enterprise Manager had a bazillion options but still felt about as flexible as a surf table.

I'm not saying it's not an viable tool, but it's not something I'll choose a database over.

Re:Cue postgres fan bois

[vadim_t]

Every DBA should have had the opportunity to experience the joy of typing "DELETE FROM Customers" and accidently hitting the [Enter] key before the typing in " WHERE Customer ID = 'Bad Customer'".

That's why first of all, you SELECT using the WHERE clause you're planning to use to check that you're choosing the right records.

If it's a very serious matter, or a multiple step operation, then wrap it all in a transaction that's rolled back at the end. Check that the right number of rows have been changed.

Inside that transaction, do a SELECT to verify that the data was changed in the right way.

Only after that, change the ROLLBACK to COMMIT, and re-run.

Re:Cue postgres fan bois

[Requiem18th]

Oh, I have to add that documenting said task objects involved inserting screenshots of the creation process, as well as a light version of how to change the server just in case the client needed to.

Re:Cue postgres fan bois

[asdf7890]

What kind of revolution are you waiting for?

Nothing myself, but the post that I was replying to stated

at least it's "remarkably changed" from Enterprise Manager

which I was disagreeing with. "Remarkable changed" implies more difference than I see over the products history while I've been using it.

Re:Cue postgres fan bois

[Blakey Rat]

Oh. Well you seemed upset that it's not "remarkably changed", but you don't know which remarkable change you're waiting for? Seems like a goofy complaint to me.

Re:Cue postgres fan bois

[asdf7890]

Oh. Well you seemed upset that it's not "remarkably changed", but you don't know which remarkable change you're waiting for? Seems like a goofy complaint to me.

Not upset, just correcting (IMO) the original sentiment.

Perhaps I chose my words badly. Wouldn't be the first time.

Re:Cue postgres fan bois

[Blakey Rat]

Well, I'm just saying if there's some specific feature you're missing, then by all means complaint about it not being there. But if you just don't like it because it's not "changed" enough, that just strikes me as making-up excuses to not like it.

Re:Cue postgres fan bois

[tweek]

Yes they did. I can only think of a handful of products that were Microsoft "originals" and even those were generic products. SQL Server is based on Sybase. In fairness, SQL Server 7 and beyond are supposed codebase rewrites.

Re:Cue postgres fan bois

[Anonymous Coward]

The fact that you refer to "Enterprise Manager" and "DTS" proves you don't know WTF you're talking about. Those two things don't even exist anymore.
SQL 2000 and down were a joke. SQL 2005 and more importantly 2008 or 2008 R2 can and will take on Oracle in just about any space. DB2 is... stagnating.

Re:Cue postgres fan bois

[Anonymous Coward]

You should have used Query Analyzer instead...

Re:Cue postgres fan bois

[Gary van der Merwe]

The old Enterprise Manager had a feature that I used to use a lot. You could open a query designer. "Design" a select query (typically add a where clause or possibly a join and a where clause on the joined table.) Execute the query, check the results. Then you could change the query type, to a delete or update statement. If you had multiple tables, it would prompt you on which table you wanted to change. It would keep the rest of the query (from, where, etc.) in tack. I can't figure out how to do this in the new Server Management Studio - frustrating.

Re:Cue postgres fan bois

[bytesex]

No, there is a difference in this example between the selected and the sorted-on field. Plus, the selected field may not be unique (which is the main reason for getting an arbitrary amount of results). If I make a webapp in which I produce a subset of one or more tables that are the result of a query that I want to present to the user in the form of pages, I may, for example select on everybody in Houston (which gives me a subset), and sort on name. If I get to use offset and limit, I can do something like this:

SELECT * FROM PEOPLE WHERE CITY='Houston' ORDER BY NAME OFFSET 0 LIMIT 20;

For the next page, I would then replace '0' with '20', to give me the next 'page' of the result set. Fantastic. However, if I did this:

SELECT * FROM PEOPLE WHERE CITY='Houston' AND ROWNUM() >= 0 AND ROWNUM() < 20 ORDER BY NAME;

I would first select the first twenty (arbitrary) results from the database, and only *then* sort them. My result could contain names beginning with 'a' through 'z' for every page ! The 'solution' proposed here, says that you should be aware of your resultset. Something like this:

SELECT * FROM PEOPLE WHERE CITY='Houston' AND UPPER(NAME) >= 'A' AND ROWNUM() < 20 ORDER BY NAME;

Assuming (and that's a big if) that the database now outputs rows in a sorted fashion (which it might do, if a proper index was used), this might work: the next page would then take the last name of the first page, and change the criterium to say: 'UPPER(NAME) >= 'BUTTHEAD'. However, it means that you can't 'go back a page', and when there are people with the same name across page boundaries, or worse: more people with the same name than fit on a single page, you're stuck. Can't go forward, can't go back.

Re:Cue postgres fan bois

[asdf7890]

OK. Let us try this in simple sentences and short words.

I was not complaining that it has not changed enough. I was not complaining about the lack of any particular feature.

I was simply disagreeing with your original assertion that it is "remarkably different" from the older tool(s). No more, no less.

Re:Cue postgres fan bois

[Cico71]

No, there is a difference in this example between the selected and the sorted-on field. Plus, the selected field may not be unique (which is the main reason for getting an arbitrary amount of results). If I make a webapp in which I produce a subset of one or more tables that are the result of a query that I want to present to the user in the form of pages, I may, for example select on everybody in Houston (which gives me a subset), and sort on name. If I get to use offset and limit, I can do something like this:

SELECT * FROM PEOPLE WHERE CITY='Houston' ORDER BY NAME OFFSET 0 LIMIT 20;

For the next page, I would then replace '0' with '20', to give me the next 'page' of the result set. Fantastic. However, if I did this:

SELECT * FROM PEOPLE WHERE CITY='Houston' AND ROWNUM() >= 0 AND ROWNUM() < 20 ORDER BY NAME;

I would first select the first twenty (arbitrary) results from the database, and only *then* sort them. My result could contain names beginning with 'a' through 'z' for every page ! The 'solution' proposed here, says that you should be aware of your resultset.

To put it simply: no.

Look: I'm not sure if we don't understand each other, if you're teasing me around for trolling purposes or if you are genuinely uneducated on the ROW_NUMBER() windowing function (which has nothing to do with Oracle's ROWNUM pseudo-column); but it doesn't work as you say.

To prove it, you just need a copy of SQL Server 2008 (express edition should do) and the last version (available on Codeplex) of the AdventureWorks2008 sample database. Then just run this query against it and play with the numbers as you want:

SELECT *
FROM (
SELECT ROW_NUMBER() OVER (ORDER BY P1.FirstName) AS rn
, P1.FirstName, P1.LastName, A1.City
FROM Person.Person AS P1
JOIN Person.BusinessEntityAddress AS T1
ON P1.BusinessEntityID = T1.BusinessEntityID
JOIN Person.Address AS A1
ON T1.AddressID = A1.AddressID
WHERE A1.City = 'Redmond'
) AS T
WHERE rn BETWEEN 1 AND 21;

Ordering is applied right after the A1.City = 'Redmond' predicate while generating row numbers with the ROW_NUMBER() function.

So

• there's absolutely no arbitrary selection for assignment of row numbers, they are assigned base on FirstName ordering
• because of this, there can't be names beginning with 'a' through 'z' for every page (unless, of course, they all fit in a single page)

I suspect you are simply wasting my time because you don't know how windowing functions work but still you are arrogant enough to try and bend Oracle's ROWNUM into a possible behavior without simply looking up some documentation. Have you *at least* seen and contemplated that the ROW_NUMBER() function comes with an OVER (PARTITION BY ..., ORDER BY ...) clause???? And that ORDER BY clause is different from the ORDER BY clause that come at the end of a SQL statement?

Let me quote from Oracle's documentation:

A ROWNUM value is assigned to a row after it passes the predicate phase of the query but before the query does any sorting or aggregation.

This is simply *NOT* how the ROW_NUMBER() windowing function works! Look it up in both Oracle's and SQL Server's documentation.

There are some additional considerations that may apply because of determinism of the sort order or duplicates because of the chosen attributes in the partitioning/ordering clause, but these are considerations that applies also to LIMIT.

Before starting again running in circles, because you have no clue about how this windowing function works,

Re:Cue postgres fan bois

[edmicman]

Too bad hitting the [Enter] key won't do a damn thing other than move to a new line? F5 to Execute your script for the win.

And anyone deleting anything from any tables should know better to double check before committing and actually executing the script. Anything else is just amateur.

Re:Cue postgres fan bois

[Anonymous Coward]

You use Enterprise Manager and Query Analyzer? A real SQL Server dba uses osql or sqlcmd at the command line. Anything else is just amateur.

Re:Cue postgres fan bois

[aevans]

login permissions aren't ANSI SQL, socket connections aren't ANSI SQL, character sets aren't ANSI SQL. There is a lot to a database that isn't covered in the ANSI SQL spec.

Re:Cue postgres fan bois

[shutdown -p now]

In this case we're talking about functionality that is covered by ANSI SQL, however. It's just provided in a way different from LIMIT, and arguably less convenient; but still, it's there.

Oracle needs to cater to business not the communit

[Anonymous Coward]

If Oracle provides what business needs/wants .. and that's what they have been doing thus far. They will be fine. Nothing to see hear move along.

Re:Oracle needs to cater to business not the commu

[houstonbofh]

Point taken... The do not need to provide the wining fork, just support it. They actually have a cost advantage if the winning fork is developed by someone else. The downside is that they can't just merge the open code back into the commercial database, and that is a significant downside.

Already Planning my project for MariaDB

[Foofoobar]

I've already started planning my open source project to support Maria. I'll probably still support MySQL but I expect the community version will have fewer options and functionality over time and will fall out of use so it's probably just easiest to start making the switch to MariaDB right now.

From my understanding, it already supports PHP and is far faster than 5.1.

Re:Already Planning my project for MariaDB

[tnk1]

I'd hope that in the short to mid term, neither side of the fork introduces something that would make either one become incompatible. That may happen eventually, but I think if one side does that too soon, it will be the end of their fork.

I think a fork here is certainly an appropriate action, but I hope that the ODA people don't go so far down the "we're not Oracle path," that they end up on the wrong side of existing users.

I personally support well over a hundred major MySQL installations right now which serve content, and my company operates hundreds more. So, obviously, any talk of divided support makes me uneasy.

In the short term, we're likely to stay with the official MySQL, but we are very interested in not getting stuck in vendor lock-in or abandonment. We definitely are trying to get away from Sybase and Oracle DBs as much as possible, so the Oracle purchase concerns me. As long as we don't have to re-code to move to Maria, it will always remain an option.

And who knows, if its faster and better, it may be a no-brainer to switch to it even if MySQL does not get marginalized or subverted. Still, that is only something that is going to happen after a full evaluation period with both sides of the fork. In the meantime, they need to be interchangeable enough so all my code on one works on the other. I cannot stress that enough.

Re:Already Planning my project for MariaDB

[Foofoobar]

Well they want to be compatible with existing MySQL versions as Monty has stated and not be a fork but they also want to fix broken functionality on the backend that SUN has been unwilling to do thus far. MariaDB from my understanding is just a community version but they would like to see their commits go up to Oracles version as well.

If they make improvements and Oracle refuses to be part of the community, then guess who is going to have the better version in the long run? I'm putting my money on that community of developers rather than the company with the overblown ego. We all know how companies with overblown ego react to the open source community after all.

Re:Already Planning my project for MariaDB

[jbolden]

It is going to be Drizzle and they are going to do it quickly. The embedded market doesn't care as much about uniform compatibility if they need performance. Maria OTOH will stay very close to MySQL legacy and the main MySQL (unless they pick up a large percentage of mindshare). So I'd expect the main MySQL and MariaDB to codeshare if both projects are doing well while expect Drizzle to fork off and never come back.

Re:Already Planning my project for MariaDB

[Ash Vince]

Personally I am more worried that a pure open source DB would be subverted by the Richard Stallman worshipers and end up restricting commercial use of the database in closed source projects through some future revision of the GNU licence.

I would love everything to be open source, but sometimes that is simply not applicable in a capitalist world where the side with the most to spend on lawyers frequently wins. I would strongly object to someone taking my code and selling a service based on my hard work just because I could not prove I was copied by them and not vice versa. Maybe although I could prove it was my hard work I could not prove it to the necessary legal standard in a court of law.

I have seen enough miscarriages of justice to know that no legal system is perfect, especially when one side is rich and the other poor.

Re:Already Planning my project for MariaDB

[derGoldstein]

Suppose this process gradually resolves itself, in any direction.
So what happens to documentation in the meantime? What happens to educational institutions and anyone who wants to learn what is generally perceived to be the number one open source DB?

This isn't an end-user software project, where "mere mortals" can try every flavor and choose whichever fits their tastes. The newly initiated will want guidance and someone to point a finger at a URL, at which they would hopefully find a big "Download Here" button and a slew of pretty PDFs they can go through, as well as references to where they should look if they want to know more, like book titles, and courses.

This isn't going to benefit the project. It seems to me that the longer this takes, the more damage will be done.

Re:Already Planning my project for MariaDB

[lamapper]

but we are very interested in not getting stuck in vendor lock-in or abandonment.

Thus you must look at the vendor that purchases and their specific history with open source projects and decide what to do.

Historically, when a proprietary vendor takes an open source project in house, especially when said open source project competes (or appears to compete) with their proprietary product, eventually they do something (or nothing, as in letting it stagnate) that results in what-use-to-be-called open source being abandoned.

If Microsoft adopts, historically they only support long enough to replace with a proprietary product and then kill it off. History is littered with examples.

If a company that has a strong relationship with Microsoft adopts, eventually the relationship results in the company either killing or abandoning the project. History gives us ample examples of this scenario as well.

The only example that I am currently aware of, and there are people who would NOT agree with me here, of an open source project going in house and continue to being supported is OpenOffice.Org (OOo) which was acquired by Sun.

Of course whenever the majority of developers jump ship after an event like this, that is usually a good tale-tale sign of the acquiring companies intentions with the open source project.

A good solid fork by the founder starts to sound safer than the official fork at that point.

I will probably continue to use the official fork for a few more months to one year to give Sun a chance to show their true intentions. As soon as something is implemented that can NOT be forked, as in they alter the license for that feature, I will move to the fork.

Equally bad, will be if the company does not dedicate enough resources to keep the official fork current, which is what I suspect is more likely given the defection of many core MySQL devs, than I will move to the fork.

Obviously I will be installing a forked version of MySQL to make sure that I can test in both the forked and official version. If for any reason the official version starts to require Java, or some other proprietary product in order to install it and/or use it, that too is a death knell for the official version.

Java can be an option, but should NEVER be required. Just my two cents.

Sun seems to love requiring Java in their apps

Re:Already Planning my project for MariaDB

[CAIMLAS]

Maybe. Maybe not. It might take a while, but unless it drags on for over a year or so, I'd wager it won't be much of an issue.

Yes, mindshare in the initiate is important. But it is not the end-all, be-all. They don't make decisions, and if they're not complete idiots, they're looking to their technical superiors for recommendations and input.

It'd take a lot of botching on both Oracle's and the "MySQL community's" part for MySQL to get seriously dropped, I think.

Re:Already Planning my project for MariaDB

[jadavis]

Well they want to be compatible with existing MySQL versions as Monty has stated

That will only last so long. First, the forks will start adding features, even if they seem like minor new options. After a short time, all of the forks have quite a few new features, and the baseline functionality is blurred with the rest of the product.

There is no "MySQL standard" like there is a SQL standard. So, the versions will just drift apart. Application programmers will now have to test on many versions and forks of MySQL to make sure it works. The mailing list for their application will be filled with "what fork of MySQL are you using?" and "Oh, you're using that one. Do you have option_only_for_forkXYZ set to foo?" and "We're about to do a release, has anyone tested on forkXZY?".

MySQL is already suffering some of that from the different storage engines (as well as many behavior-changing configurable options). Multiplying the complexity by 2 or 3 viable forks will mean that most combinations of MySQL will be far less common than PostgreSQL/FirebirdSQL.

My prediction is that people will get excited about the new forks, but the application support will never really match it. People will use it for a few internal projects that only support that specific fork. Ultimately, it will just lead to frustration. I do not think that anything except mainline MySQL will really be a serious player.

These forks may be interesting as a playground for new ideas, however -- and maybe that's how the authors intend them.

Re:Already Planning my project for MariaDB

[jadavis]

It might take a while, but unless it drags on for over a year or so, I'd wager it won't be much of an issue.

I think all of the forks will be in a state of flux for several years. Why would you expect it to be less than a year? Most database projects couldn't do more than one release in that time.

Fork it

[sakdoctor]

If I could only think up a cliched pun using the term forked, I would be sure to get the converted +5 funny mod.

If only.

Re:Fork it

[jgtg32a]

but +5 funny doesn't get you mod points

Re:Fork it

[Fantom42]

If only I could think up a cliched self-referential variation on your posting to correct your English usage, I would be sure to get the coveted +5 interesting mod.

If only.

Re:Fork it

[Anonymous Coward]

If I could only think up a cliched pun using the term forked, I would be sure to get the converted +5 funny mod.

If only.

fork + shoe = spleen

Re:Fork it

[amicusNYCL]

I didn't realize that any mod got you mod points.

Funny does nothing for your karma, though.

Re:Fork it

[houstonbofh]

but +5 funny doesn't get you mod points

But it does give you a nice achievement. :)

Re:Fork it

[value_added]

If I could only think up a cliched pun using the term forked ...

You're way ahead of me. I'm stuck at figuring out what "wrong side of the fork" means. Where's BadAnalogyGuy when you need him?

Re:Fork it

[Locke2005]

You mean like "Those MySQL users are really forked now that Oracle has control of MySQL!"?

Re:Fork it

[Anonymous Coward]

I think they meant the "dark side" of the fork. Whatever that is.

Re:Fork it

[fm6]

Yeah, it's forked up.

Re:Fork it

[h4rm0ny]

I got a +5 funny post once.

When I was being serious. :(

Re:Fork it

[EvanED]

The line between humor and seriousness can be very thin at times.

One of my favorite quotes about gay marriage was said by Jon Stewart during an interview of him, which was something like "At first I was really against it. Because I love my wife. And then I realized that it wouldn't be mandatory."

Re:Fork it

[derGoldstein]

Funny does nothing for your karma, though.

Hogwash.
There are piles of research papers proving that Funny does wonders for your karma. Especially if you combine it with Torgo's Executive Powder.

Re:Fork it

[derGoldstein]

You're way ahead of me. I'm stuck at figuring out what "wrong side of the fork" means

It's the business end. Whenever it's pointing at you.

Re:Fork it

[dgatwood]

Forked upstream beyond all repair.

Re:Fork it

[derGoldstein]

Just once? I usually get modded "funny" when I'm serious, and "informative" when I'm joking.

Most /. mods are very bored individuals. It gives them something to do. They're not like us, the /. posters, who are intelligent, thoughtful, handsome, and socially adept.

Re:Fork it

[derGoldstein]

In soviet Russia, the project forks YOU!

Re:Fork it

[fm6]

What an uneuphonious comment. You must have a tine ear.

Re:Fork it

[kusanagi374]

So, I guess you were joking right now, eh?

Re:Fork it

[zonker]

MySQL is killing itself from the inside like a flan in the cupboard. Oracle hasn't even gotten its hands on it yet.

Re:Fork it

[TheMuffster]

Thank god you got modded "imformative"... Otherwise I might have not recognized you're just being funny...

Re:Fork it

[Anonymous Coward]

You forgot "independent".

(as in, independent of their mom)

Re:Fork it

[Flammon]

And you get a +5 Informative for that comment. Oh the irony.

Re:Fork it

[iapetus]

Inforkmative, surely?

Re:Fork it

[nurb432]

You have been sporked?

Re:Oracle needs to cater to business not the commu

[Anonymous Coward]

Funny you should say that. We're working on dumping Oracle licenses where they're not needed in favor of MySQL. That was going to be expanding our Sun support contract. Oddly enough - it might end up expanding our Oracle contract. That is, unless Oracle gives us reason to look elsewhere for MySQL support.

Knowing Oracle...

[MikeRT]

I really don't think they care, nor do I think that they bought Sun for MySQL. I wouldn't be surprised in the least if they kick it to the curb here and either create a lighter low end Oracle DB for MySQL's users or if they revamp Sun's support for PostgreSQL by making it the preferred entry-level compliment to Oracle DB. Oracle already has a huge service wing, and using PostgreSQL wouldn't hurt them in that capacity as a vehicle for selling Oracle DB where it makes sense to those customers.

Re:Knowing Oracle...

[Reality Master 201]

Yes, exactly. And PostgreSQL's capabilities and feature set is already closer to Oracle than MySQL's is or is likely to be in the future.

Re:Knowing Oracle...

[h4rm0ny]

That's an interesting point of view. I saw things slightly differently, so say what you think of my take on it. I agree that Oracle might not actually care so very much that they got a free MySQL when they bought SUN. But in so far as they do pay attention things, I would think they'd steer people away from PostgreSQL toward MySQL rather than the other way around. My reasoning is that there is a clear and significant gap between MySQL and Oracle DB. MySQL is never going to draw many customers away from using Oracle DB, and mostly the other way around as well. But I see the gap between PostgreSQL and Oracle as being much smaller, both in the capabilities of the two databases and the easier time you have moving from one to the other (in either direction). I think that makes PostgreSQL more of a threat to the Oracle install base and thus something they would prefer to keep people away from. Using MySQL as a stalking horse makes more sense to me. Thoughts? Matches the real world or too conspiracy?

Re:Knowing Oracle...

[moderatorrater]

Considering that Oracle acquired InnoDB a few years ago with a company for whom that was the only real asset, I think it'd be a mistake to think that Oracle doesn't have plans for MySQL. InnoDB is the engine for those who are moving in the direction of more robust database solutions, and if you're supporting people on InnoDB and MySQL who could use a feature in Oracle, it's mighty easy to train tech support to say, "Well, there's a horrendous workaround, or you could upgrade to Oracle which will fix all your problems. We'll even send a tech out there to help you migrate, which we've gotten pretty good at since we've got experts for both codebases in our company."

Re:Knowing Oracle...

[jbolden]

Postgres is getting much better speed wise. But Oracle's main selling point vs. DB2 traditionally is how fast it is. MySQL is also fast comparable to Oracle. This is less of an issue than it was a dozen years ago but I'm not sure how much that mindset still survives at Oracle.

Re:Knowing Oracle...

[Reality Master 201]

But Oracle's main selling point vs. DB2 traditionally is how fast it is.

For amusement, make that point in front of a bunch of IBMers/DB2 consultants. If you're lucky, you'll hear how Oracle's SQL optimizer is the one IBM rejected in favor of the one used in DB2.

Gotta say, though, I do prefer DB2 to Oracle.

Re:Knowing Oracle...

[jbolden]

Interesting where do you see the comparative area of strength.

I'm assuming something other than mainframe intergration / COBOL integration where DB2 is really really strong.

Re:Knowing Oracle...

[raftpeople]

There is a lot I like about Oracle, but one thing that just bugged the heck out of me was how poor the optimizer was. I was used to the as400 database (they kind of call it DB2 but not sure how related it is to DB2 on the other platforms) and the as400 optimizer seemed to be pretty damn smart. I was surprised I had to invest so much time with hints or restructuring the sql to get good performance with Oracle in some situations. But, I did like the performance when it wasn't confused and I really liked the read consistency and multi-versioning stuff.

Re:Knowing Oracle...

[turbidostato]

"My reasoning is that there is a clear and significant gap between MySQL and Oracle DB. MySQL is never going to draw many customers away from using Oracle DB, and mostly the other way around as well. But I see the gap between PostgreSQL and Oracle as being much smaller"

And that's exactly why Oracle could support PostgreSQL better than MySQL: it's all about bussiness opportunities. Postgres is much more a "typical" RDBM while MySQL tends to do thinks their own way*1, so look at the outcomes:
1) You are using PostgreSQL very publicly backed up by Oracle and you (or your boss) think you need "more". The natural choice would be 10 out of 10 time going for Oracle's big product which is like PostgreSQL, only more.
2) You are using MySQL very publicly backed up by Oracle and you (or your boss) think you need "more" but then find going for "more" means changing quite a lot of things (new or heavily retrained DBAs, revise tons of code, etc.) then, while probably your first choice still would be stick to Oracle, there's a non zero chance that you go out to study the market and go with anyone else.

*1 I don't think you can find a professional DBA with experience on Oracle, DB2, Sybase, etc. and then PostgreSQL and MySQL that would choose MySQL over PostgreSQL not even offering him money.

Re:Knowing Oracle...

[saleenS281]

They already do, it's called oracle express.

What the fork?

[Anonymous Coward]

Just tell me which one to use for the salad, so my ignorance doesn't trip me up like it did with that bidet that one time.

Re:What the fork?

[osvenskan]

You used your bidet fork for the salad?

Re:What the fork?

[Mr. Underbridge]

Just tell me which one to use for the salad, so my ignorance doesn't trip me up like it did with that bidet that one time.

Are you talking about those flavored water fountains they have in bathrooms in France?

Re:What the fork?

[Anonymous Coward]

Yeah, I know, you're not supposed to drink from it.

Not as serious...

[Anonymous Coward]

This really isn't as serious as is implied. MySQL is GPL. The forks are GPL. Therefore, if Oracle wishes, they can just cherrypick the best patches from all forks and integrate them back into their codebase, community involvement or not. I expect MySQL to remain the official MySQL, unless it completely stagnates, simply due to name recognition if nothing else.

Re:Not as serious...

[dvice_null]

> The forks are GPL. Therefore, if Oracle wishes, they can just cherrypick the best patches from all forks

They can. But then they can't dual license it anymore as they don't own copyright for the whole source.

Re:Not as serious...

[harryandthehenderson]

But they can't integrate it back into a proprietary commercial product like MySQL AB or Sun could do if they don't hold the copyrights to the contributed code to the forks. Oracle isn't stupid like Sun and just gives shit away. If they can't integrate the work back into a sellable commercial product they just aren't going to bother.

Re:Not as serious...

[Locke2005]

You are assuming Oracle want MySQL to continue to thrive. Personally, I believe the bought Sun just to get control of MySQL so they could kill it, since it was taking sales away from them. But this won't be totally effective; we need to get all the MySQL users to rally behind one of the forks not controlled by Oracle.

Re:Not as serious...

[Anonymous Coward]

why can't they just study the patches, and then rewrite them? Copyright problem solved. Unless you patent the patch, you can't stop this. And then, the patch wouldn't be GPL'ed either.

This isn't unethical either. Its one of the points of GPL that the source is available for all to study and learn from...

Re:Not as serious...

[Your.Master]

Doesn't work so neatly as that. GPL is a copyright and you can infringe copyright without using literally identical material.

I'm not so sure the ethical argument works out either. Just because some GPL advocates say they want others to study and learn from it, doesn't mean that the people who actually contributed to it intended for it to be studied, learned from, and then essentially plagiarized without the license. I doubt even most advocates who speak those words are intending that you do this.

Re:Not as serious...

[h4rm0ny]

why can't they just study the patches, and then rewrite them? Copyright problem solved. Unless you patent the patch, you can't stop this. And then, the patch wouldn't be GPL'ed either.

This isn't unethical either. Its one of the points of GPL that the source is available for all to study and learn from...

GPL 3 would stop the patent issue. Subject to testing in court. And subject to software patents being mostly US-specific. And subject to the code base moving forward enough under GPL3 to make backwards compatability with a GPL2 fork no longer trivial to preserve. But that's just an aside to an argument that will remain hypothetical forever anyway as I can't see any main MySQL branch moving to it.

The real issues are simply how much work would be involved in re-writing the code and whether your version was different enough that you could defend it as not being a disguised copy paste. I think the latter would never actually arise, but the former would be a nuisance. Not a showstopper - I think you're right this could be done, but it would be annoying and become more annoying as time went on and the forks grew further apart. Any Open Source features of significance that Oracle cares about are probably going to be quite large ones to re-implement. The big one would be the Falcon table type.

Re:Not as serious...

[idontgno]

Personally, I believe the bought Sun just to get control of MySQL so they could kill it, since it was taking sales away from them.

A cogent and insightful theory, except that it's precisely like Mack buying out Segway because it was taking truck sales away from them.* Oracle doesn't have a credible product (or even a foot in the door) in any market where MySQL is already effective.

But this won't be totally effective

For the basic reason I stated above

we need to get all the MySQL users to rally behind one of the forks not controlled by Oracle.

Fair enough; I dislike Oracle's business dealings and basic arrogance enough to get me behind a genuine grassroots fork. But don't succumb to Amiga Persecution Complex. If it happens, the non-Oracle fork's ultimate poor uptake and small community won't be because of some sinister Oracle Trilateral Commission keeping the little guy down; it'll be the little guy just being little.

*I'd like to point out the nifty Slashdot-compliant car analogy. OK, technically, truck analogy.

Re:Not as serious...

[Alpha830RulZ]

Oracle doesn't have a credible product (or even a foot in the door) in any market where MySQL is already effective.

What about Oracle Express?

We've used MySQL, Oracle Express, and SQL Server express for low end customer applications, where we really only need a few hundred megabytes of data for our application. My order of preference is SQL Server, MySQL, and Oracle.

Re:Not as serious...

[Anonymous Coward]

Oracle isn't stupid like Sun

LMAO
Oracle is pretty stupid. Instead of releasing a product that works, they just lie. Sun had some serious talent, put out good product eventually (Solaris 10). Its really too bad what's happened. I'd distance myself from both now.

Re:Not as serious...

[zefrer]

They can dual license it if the code author allows them to. It is common practice in these dual license scenarios to ask for the code author to do this. Whether they allow it or not is another matter entirely.

Also note that most countries have a minimum size (not well defined obviously) that a work must be in order to be considered copyright-worthy. A one byte patch for example would not be covered by copyright.

Re:Not as serious...

[idontgno]

Well, I did say "credible". If MySQL is a Segway, Oracle Express is a tricycle. (Not the cool kind, or even the useful kind. The pre-school kind.)

Justferinstance, if you need over 4Gb of tablespace in MySQL, it's just a couple of

ALTER TABLE

statements away. If you need over 4Gb of tablespace in OE... your Oracle account exec will be happy to help you. Bring your wallet.

Re:Not as serious...

[Alpha830RulZ]

Yeah, but there are a ton of apps that will never use 4 GB. For those, Oracle Express, or SQL Server Express are perfectly usable. Though I wouldn't willingly use Oracle by preference; I hate Oracle. I hate the company, the dev tools, and the complexity of the environment.

I use and like MySQL. I'm not dissing it in any particular way, though I wouldn't use it for a real corporate main data store because of its weak RI features. It works great for a lot of lighter weight applications, and we use it daily for reporting databases.

As to your tortured analogy, I think it would be more accurate to say the OE is a tricycle on a leash.

This need not be Bad News.

[anonymousNR]

Really this need not bad news.
This is all part of an opensource product life cycle.
It had its peak time, still being used in some great projects.
In good hands would still have a great future.

Re:This need not be Bad News.

[Anonymous Coward]

In good hands would still have a great future.

And then Oracle bought it and shot that dream right in the ass. Used an Oracle product lately? They freaking SUCK. I don't imagine buying out Sun will suddenly shoot a magic bea

Re:This need not be Bad News.

[Anonymous Coward]

This is the most poetic comment I have ever read on Slashdot. It calmed me.

Re:This need not be Bad News.

[ScrewMaster]

In good hands would still have a great future.

I understand that Allstate is interested.

The obvious solution

[Anonymous Coward]

Too many forks? Just create another fork.

What usually happens is a fork progresses to the point where all the developers want to re-write all the code from scratch anyway. Just skip to that.

Re:The obvious solution

[harryandthehenderson]

Rewriting codebases unnecessarily. The classic syndrome of open sores software!

Re:The obvious solution

[larry bagina]

I'd say open sores has the opposite problem. If something is "good enough", people use that (either the app or the code) instead of writing new, better code. Times change. Faster CPUs, more memory, multiple processors, threading, improved libraries, better algorithms, better security practices, better understanding of end requirements, etc.

Re:The obvious solution

[ScrewMaster]

Too many forks? Just create another fork.

Blame Larry. As usual, Mr. Ellison came along and stuck a fork in it.

One obvious question

[jd]

How much of MySQL can be split off entirely and developed as a "common" frontend or a "common" backend? If there's a specific module or layer that everyone can more-or-less agree on in terms of code and API, then make that your "MySQL" and make the rest a dependency where the variations are interchangeable.

The benefit of this kind of approach is that you can have something that can be highly tuned and yet benefit from the large existing programmer base. But it only works if you can split the code meaningfully and keep the interface across the split stable.

(GCC is close-ish to this sort of idea - anyone can fork a frontend or develop a whole new one without colliding with the backend developers, but the interface is far more fluid. Which is fair enough for a compiler, but won't work if you're talking databases. glibc/newlib/ulibc could also be considered similar in that they're fairly interchangeable but that's a special case.)

PostgreSQL: Why don't people use it that much?

[bogaboga]

I have always wondered why people do not use PostgreSQL that much. It is better than MySQL in terms of stability and scalability.

You might wonder how I came to this conclusion. Well, I have used MySQL with MythTV and I have gotten sick and tired of corrupted databases/tables.

I have a read a reviews of PostgreSQL's stability and scalability beyond two cores and have no doubts it is better than MySQL on this front, though there have also been crowds here at Slashdot who think MySQL is better. My experience suggests otherwise.

Re:PostgreSQL: Why don't people use it that much?

[diegocgteleline.es]

Postgresql has a terrible marketing and image, mysql doesn't.

Re:PostgreSQL: Why don't people use it that much?

[Anonymous Coward]

Damn, you have done a LOT of testing to come to your conclusion. Do you have a blog? if so, I would like to subscribe.

Re:PostgreSQL: Why don't people use it that much?

[Anonymous Coward]

Religious debate. Way back in the Olden Days, postgres was an unstable mess. Absolutely godawful. Corrupted databases, lost data, slow as moleasses, constant need to restore from backups. Completely worthless.

At the same time, MySQL was fast, stable, fast, worked well enough, slightly feature incomplete, and fast.

Those old stereotypes have stuck. MySQL nowadays? Actually pretty feature complete. PostgreSQL nowadays? Pretty stable and solid.

But nobody updates their perceptions, so the old ideas shine through.

Re:PostgreSQL: Why don't people use it that much?

[chabotc]

Anecdote != fact

A very important thing to keep in mind. For instance for you 'databases and tables corrupted', for thousands of others though this never occurred.

Re:PostgreSQL: Why don't people use it that much?

[Anonymous Coward]

It's true MySQL has a cute name, whereas nobody is really sure how to pronounce PostgreSQL and their icon is a big elephant. Well makes sense for a database system I guess...

Re:PostgreSQL: Why don't people use it that much?

[jd]

Well, yes, PostgreSQL is good. So is Ingres, now that there's a GPL version. Even fewer people use that. (Hell, I don't even know a distro that has Ingres packages.)

And, of course, those are only a few of the database engines out there. When you consider the sheer number of different types of database there are (hierarchical, relational, OLAP, object-oriented, indexed sequential flatfile, random access sequential, and so on), it's obvious that there's a lot of room for specialist engines.

Chances are, though, that users don't have meaningful access to even a fraction of the engines out there, or any sensible way to compare what their requirements are with the types of engines they can choose between.

Open Source is all about choice, but without the information or the means to make that choice, the choice is essentially an empty one.

Re:PostgreSQL: Why don't people use it that much?

[langelgjm]

My, what extensive experience you having using MySQL!

Seriously though, the MySQL databases in MythTV do seem to get corrupted a lot. I use KnoppMyth, so the fix is usually just a simple "optimize_db", but it happens often enough that I've thought of automating the process.

Re:PostgreSQL: Why don't people use it that much?

[harryandthehenderson]

Anecdote != fact

Unless he is making it up his anecdote would be a fact. You seem to have confused this with saying that anecdotal evidence is not generalized scientific evidence.

Re:PostgreSQL: Why don't people use it that much?

[Anonymous Coward]

PostgreSQL is quite a nice database.

It's a shame it has such a crappy SQL parser and contains only overthought, overblown versions of otherwise simple everyday functions.

Seriously, I have to use "QuotedIdentifiers"? Bite me. Even MS SQL Server doesn't require delimiters. And in the few cases where you get an ambiguous reference, it tells you that in plain english. And then it allows you to use "Quotes" or [Brackets]. Or you can sometimes just specify the object hierarchy. (Example: DatabaseName.OwnerName.TableName.ColumnName won't return an error even if just ColumnName will.)

And then there's the increment functions. MySQL is a bit lacking, with only AUTO_INCREMENT and LAST_INSERT_ID. SQL Server has a similar setup to MySQL using IDENTITY(start, step) and SCOPE_IDENTITY(). But PostgreSQL uses that atrocious monstrosity of SERIAL data types and a damned Sequence object. And heaven help you if the Sequence gets out of sync. Your data is fucked at that point. The DB engine won't fix it. You aren't allowed to fix it. And the next time your app tries to use it, there's a good likelihood that it will lose data.

MySQL using InnoDB is tolerable. PostgreSQL is safer. But to be perfectly honest, I'm growing to actually like MS SQL Server. It's as simple and flexible as MySQL, and has the data integrity of PostgreSQL. Oh, and it makes stored procedures easy too. All at the same price as either of them. (OK, it's free-as-in-beer for the Express edition. It's under $50 for the "developer" edition, which is a full enterprise license with unenforced restrictions. Spending thousands on SQL Server is for suckers.)

Re:PostgreSQL: Why don't people use it that much?

[Anonymous Coward]

In my case the reason is real-time replication to remote slaves, which seems a lot easier with mysql than with postgres

Re:PostgreSQL: Why don't people use it that much?

[K. S. Kyosuke]

I am not certain whether any form of "feature complete" counts when "final releases" are filled with bugs. MySQL might have been stable when its feature set was next to nil, but I will sooner return to FoxBase than to prefer MySQL 5. over Firebird (or PostgreSQL). PostgreSQL is has been actually *designed*. And it shows. Without good design, sustained performance improvements such as these and feature improvements such as these are unthinkable.

Re:PostgreSQL: Why don't people use it that much?

[vadim_t]

And then there's the increment functions. MySQL is a bit lacking, with only AUTO_INCREMENT and LAST_INSERT_ID. SQL Server has a similar setup to MySQL using IDENTITY(start, step) and SCOPE_IDENTITY(). But PostgreSQL uses that atrocious monstrosity of SERIAL data types and a damned Sequence object. And heaven help you if the Sequence gets out of sync. Your data is fucked at that point. The DB engine won't fix it. You aren't allowed to fix it. And the next time your app tries to use it, there's a good likelihood that it will lose data.

There's no SERIAL datatype. It's simply a shortcut for creating a sequence, and is equivalent to saying "integer DEFAULT nextval('tablename_colname_seq') NOT NULL".

Also, I'm feeling really curious about this, because:

1. Why would the sequence get out of sync in the first place?
2. If it does, why doesn't using nextval and setval work for fixing it?
3. How will the app lose data? If somehow it generates the same number twice, any attempt to INSERT a row would simply fail.

So please explain how you get what you said to happen.

Re:PostgreSQL: Why don't people use it that much?

[larry bagina]

debian has packages for firebird. Two in fact, classic (uses separate processes) and superserver (uses threads).

Re:PostgreSQL: Why don't people use it that much?

[mzs]

Also indexing was REALLY slow and to backup you needed to take the db offline early on.

Re:PostgreSQL: Why don't people use it that much?

[ppanon]

Yeah, having Dick Cheney as a spokesman really hurt them. Potential customers confuse "extensible data types for queries" with "enhanced interrogation techniques".

Re:PostgreSQL: Why don't people use it that much?

[Anonymous Coward]

I have always wondered why people do not use PostgreSQL that much.

Uh? Because PostgreSQL is a good SQL-DBMS implementation and a lot of people simply don't grasp SQL and DBMS technology in general?

Re:PostgreSQL: Why don't people use it that much?

[Anonymous Coward]

MySQL is still significantly more noob-friendly too.

Re:PostgreSQL: Why don't people use it that much?

[Tablizer]

Postgresql has a terrible marketing and image, mysql doesn't.

Please find a better name for it. Have a contest. And it's hard to pronounce, and hard to remember unless you think "post-nasal drip" (the only "catchy" memory phrase I can think of), which is NOT a positive association.
       

Re:PostgreSQL: Why don't people use it that much?

[ShieldW0lf]

Skype uses Postgres. That's a pretty glowing endorsement.

Re:PostgreSQL: Why don't people use it that much?

[Anonymous Coward]

You might wonder how I came to this conclusion. Well, I have used MySQL with MythTV and I have gotten sick and tired of corrupted databases/tables.

So based on one experience with MythTV, you're an expert on mysql scalability and stability?

Re:PostgreSQL: Why don't people use it that much?

[Anonymous Coward]

Licence pedantry is why, IMHO.

Think about it as analgous to BSD vs Linux. They both do roughly the same thing in roughly the same way. One has been around longer, is more mature, and is in many ways better than the other. The other simply has the limelight and, for many people, a preferable licence, which is why mysql is/was/tends-to-be bundled in with distros.

People get their dangfangled lunix thing, mysql comes with it and they run with it without knowing any better. So they are either innocently unaware that something more suited to their needs could be out there, or they could be actively ignorant e.g. "it's not GPL so it's not free, I have to use GPL to maintain my liberty!!!" and other RMS vomit. People who actually understand what each system is capable of and make an educated choice either way are a rare breed.

My current gripe with MySQL goes with having company staff using a bugtracker tool that suffers the same "copying to tmp table" lockup problem mentioned in this thread, which has been open since 2005. The MySQL devs response can be summed us as "meh... the fix is in 6 alpha, so kiss my(sql) ass". Well, sorry chaps, but we don't run production systems on alpha software for a reason. If the fix we've been waiting nearly four years for isn't in apt, then I suggest you get it there.

My company does not want to migrate to another DBMS, and they don't particularly want to fix the queries that are triggering this behaviour from mysql (nor should they have to!), so I've spent the last few months tweaking and retweaking mysql's configuration with marginal gain. Now the company has dished out for a brand spanking new box with a trillion cores and a bucket of ram, despite me pointing out that MySQL likely wouldn't be able to effectively use all the resources (see links above for sustained throughput and multi-thread I/O graphs). So because the MySQL devs refuse to fix their fuck up in a timely manner, I have to sit here and get irritated while my employer makes Dilbertonian decisions. Thanks, MySQL!

Re:PostgreSQL: Why don't people use it that much?

[Anonymous Coward]

When were these olden days ?

I started using PostgreSQL seriously in version 6.x with no troubles at all.

I was forced to use MySQL 3.x to support some OSS products and it was such absolute crap of a self corrupting database that it soured me against itself forever. It didn't evan gain any useful features until version 5 or 5.1.

Re:PostgreSQL: Why don't people use it that much?

[DarkOx]

Spending thousands on SQL Server is for suckers

I agree most applications would be well suited to PostgreSQL, MySQL, or possibly Sybase if you need TSQL for legacy reasons and something bigger then the Express/Desktop Editions of SQL Server.

If however you were implying as you seemed to be that you one should just strain the licenses terms of SQL Server you are out of your mind. Why would a business risk getting itself sued into oblivion just to save a couple grand on SQL licensing?

Re:PostgreSQL: Why don't people use it that much?

[williamhb]

Postgresql has a terrible marketing and image, mysql doesn't.

Everyone knows that postgresql is hiding a monstrous secret.

Re:PostgreSQL: Why don't people use it that much?

[Anonymous Coward]

In an alternate universe, everybody uses postgres so of course you're pimping mysql.

Re:PostgreSQL: Why don't people use it that much?

[Ragica]

Funny, I just got back from the first day of PgCon 2009 (in Ottawa). One of the presentations I went to today was regarding OpenStreetMap's switch from MySql to PostgreSQL last month. I think theywould beg to differ about feature completeness. It was kind of sad looking at their schema: one could really sense the limitations of mysql they had to design around. Ouch. Apparently in the end it was lack of MyISAM transactions causing constant problems with their volume of updates combined with InnoDB's lack of text search (can't have your cake and eat it too, apparently) that pushed them over the edge.

Re:PostgreSQL: Why don't people use it that much?

[ModernGeek]

Just like the, "Apple is dying, their hardware is expensive, and microsoft owns them now." crowd.

Re:PostgreSQL: Why don't people use it that much?

[Blakey Rat]

Because web hosting companies don't install it. They install MySQL, or MS SQL (if they're .net hosts.) That's it.

Re:PostgreSQL: Why don't people use it that much?

[shutdown -p now]

And then there's the increment functions. MySQL is a bit lacking, with only AUTO_INCREMENT and LAST_INSERT_ID. SQL Server has a similar setup to MySQL using IDENTITY(start, step) and SCOPE_IDENTITY(). But PostgreSQL uses that atrocious monstrosity of SERIAL data types and a damned Sequence object.

PostgreSQL seem to have copied that bit from Oracle, which also uses sequences. But sequences are actually superior, and here's why.

With sequences, you can emulate auto-increment behavior easily (you use the NEXTVAL or whatever the syntax in a specific implementation is, right in your INSERT). On the other hand, with sequences, you can generate the IDs first, and then generate a bunch of INSERTs with those IDs, with only two roundtrips: one for SELECT NEXTVAL - for as many records as you want - and another for multi-record INSERT. Why would you want a bunch of IDs? When you are batch-inserting records into a bunch of tables that reference each other (e.g. 10 records into table A, each of which has 10 new child rows in table B, each of which has 10 more new rows in table C). In MSSQL with auto-generated keys, you have to go about it that way:

1. INSERT records into A and retrieve the generated keys.
2. Use retrieved keys to generate INSERTs for new records in B; retrieve generated keys.
3. Use retrieved keys to generate INSERTs for new records in C.

For more tables in the hierarchy, you do more roundtrips. Now the same with sequences:

1. Request generation of keys for all records you'll be inserting, and retrieve generated keys.
2. Generate INSERTs for all new records using retrieved keys.

And it doesn't matter how deep the tree is, it's only 2 batches.

Of course, for MSSQL you can do away with generated keys completely, and use GUIDs, which can be generated on the client in the first place. But that makes indexes that much bigger, and queries that much slower...

nd heaven help you if the Sequence gets out of sync

This bit I don't understand at all. What do you mean by "sequence gets out of sync"? It's not really in sync with anything, it's just a simple counter that's guaranteed to safely produce unique incrementing values even for concurrent requests.

Re:PostgreSQL: Why don't people use it that much?

[kriston]

But only one of the two can claim both SQL86 and SQL99 compliance.

Re:PostgreSQL: Why don't people use it that much?

[NoelProf]

When were those olden days? I've been using PostgreSQL in production since 2001 because MySQL was (and is) relatively unstable and has an incomplete set of SQL features. Postgres has always been way more stable than MySQL. IMHO, PG is also way more straightforward than MySQL, MS SQL and Oracle. MySQL in particular has loads of hidden tricks and gotcha's. For example, referential integrity.

Re:PostgreSQL: Why don't people use it that much?

[Anonymous Coward]

Actually not really true. Postgres in the early days always was way more stable than MySQL, the speed however was less, because it always supported transactions, while MySQL started as an ISAM system with some sqlish language on top of that. But I never ever recall having the problem of corrupted data on postgres not even in version 6. The main issue was simply, that there was no decent windows port and postgres had to rely on Cygwin which made the db slow und unstable while it was not!

Add to that that early PHP versions were hardcoded against MySQL, and PHP as well got loads of momentum!

Things have changed by now. While the MySQL quality still is questionable, PostgresSQL has accellerated tremendously and also has a native windows port and good installation and UI maintenance tools but the problem persists while being leagues better, and believe me I have seen big installations where postgres chugs along happily never failing their users, MySQL still has the mindshare.

If you are looking for a real enterprise ready db in the OSS area PostgreSQL is the way to go!

Re:PostgreSQL: Why don't people use it that much?

[richlv]

because postgres people come on every goddamn article about mysql and post things like that.
so mysql users decide that all pgsql users are awful and run away from pgsql.
see, that is the reason.

Re:PostgreSQL: Why don't people use it that much?

[Splab]

GP is bit off his marbles, but your sequence can get "out of sync" if a transaction takes a value, makes an insert but decides to rollback - this means there will be a hole in the sequences issued. That is unless you define your sequence as dense, in which case the sequence id will be available again (do note however, using dense sequences means only one insert transaction can be open using that sequence at a time.)

At no point are you risking losing data over a sequence object, unless your application is really retarded and decides to guess what sequence is being issued for it.

Re:PostgreSQL: Why don't people use it that much?

[bytesex]

Are you trolling ? You don't have to quote your identifiers if they don't contain spaces in postgres. And if you don't know what sequences do or why they're useful, then you probably don't use any normalization anyway and you're better off using BerkeleyDB. I suppose your sequence gets 'out of sync' after an import (which doesn't include setting the sequence), and that by 'losing data' you mean that your inserts start failing on your unique constraints. This is why you reset your sequence after an import. If it's the 'holes' in your numbering that you're concerned about, then you haven't understood what primary keys are for. You don't count() with them, for example.

Re:PostgreSQL: Why don't people use it that much?

[Thundersnatch]

Of course, for MSSQL you can do away with generated keys completely, and use GUIDs, which can be generated on the client in the first place. But that makes indexes that much bigger, and queries that much slower...

We were worried about that once upon a time, but it turns out to have been a case of preemtime optimisation. We benchmarked one of our prototype apps with both INTEGER and UNIQUEIDENTIFIER primary keys. There was no real difference in performance (likely because it was operations on the other columns that actually took up all the time, not index seeks).

The real thing to watch out for with GUIDs as primary keys is index fragmentation... inserts are spread randomly throughout the table if you cluster on the GUID. This, in turn, causes your DB to use about twice as much space as it would with sequential keys unless you rebuild clustered indexes frequently. Our solution was a compromise: use GUIDs as primary keys (and get all the of the attendant advantages, especially multi-master replication). We also create a clustered index on some other column in the table that makes more sense for the application, such as a date/time value, or an appropriate string value that gets searched a lot.

Re:PostgreSQL: Why don't people use it that much?

[Dan Ost]

Are they happy with the switch to PostgreSQL?

Any unforeseen gotchas?

Re:PostgreSQL: Why don't people use it that much?

[Dan Ost]

In the past, people have posted web hosting companies that do allow you to use PostgreSQL.

Perhaps some nice karma seeker will find one of those posts and repost that information.

Re:PostgreSQL: Why don't people use it that much?

[fulldecent]

Maybe it's because LAPP sounds stupid

Re:PostgreSQL: Why don't people use it that much?

[jadavis]

I think the #1 surprise to people coming from MySQL is that PostgreSQL is more likely to throw errors for poorly-written queries or constraint violations.

Download the 8.4 beta, install it, run an application, and if you see any more surprises or gotchas, beta time is a great time to point them out ;)

Re:PostgreSQL: Why don't people use it that much?

[jadavis]

Seriously, I have to use "QuotedIdentifiers"?

Most people just use unquoted identifiers, which are folded to lower case, and seems pretty simple to me. I believe PostgreSQL follows the standard here, except that it folds to lower case (for readability) while the standard folds to upper case.

And heaven help you if the Sequence gets out of sync. Your data is fucked at that point. The DB engine won't fix it. You aren't allowed to fix it. And the next time your app tries to use it, there's a good likelihood that it will lose data.

Can you describe this situation? The only thing I'm aware of is if you mess up a restore, and end up with data in the table but the sequence is starting at 1. But that's really easy to fix, and the only problem I can see is that you'd get unique constraint violations.

Re:PostgreSQL: Why don't people use it that much?

[Just Some Guy]

GP is bit off his marbles, but your sequence can get "out of sync" if a transaction takes a value, makes an insert but decides to rollback - this means there will be a hole in the sequences issued.

Is that ever a problem?

Re:PostgreSQL: Why don't people use it that much?

[shutdown -p now]

The real thing to watch out for with GUIDs as primary keys is index fragmentation... inserts are spread randomly throughout the table if you cluster on the GUID. This, in turn, causes your DB to use about twice as much space as it would with sequential keys unless you rebuild clustered indexes frequently. Our solution was a compromise: use GUIDs as primary keys (and get all the of the attendant advantages, especially multi-master replication). We also create a clustered index on some other column in the table that makes more sense for the application, such as a date/time value, or an appropriate string value that gets searched a lot.

Or you could just use a GUID-generating algorithm that produces identifiers which are monotonically increasing (yes, it is possible) - MSSQL even has a special function for that, NEWSEQUENTIALID(), and of course the same thing can be done on client side as well.

Re:PostgreSQL: Why don't people use it that much?

[Anonymous Coward]

It's under $50 for the "developer" edition, which is a full enterprise license with unenforced restrictions. Spending thousands on SQL Server is for suckers.)

Or people who like to use legitimate software.

Re:PostgreSQL: Why don't people use it that much?

[greg1104]

There aren't as many hosting companies that support PostgreSQL, but they are out there. http://www.postgresql.org/support/professional_hosting has a list.

Re:PostgreSQL: Why don't people use it that much?

[fatp]

Actually I've faced "sequence gets out of sync" in Oracle 8i once upon a time. The database crashed and the sysadmin recovered it. Afterwards, when a sequence was used, and ORA-600 error occurred. When preparing for submitting a service request, I noted that one parameter of the ORA-600 was incrementing each time I used the sequence. I then compared that value with the data generated with the sequence, and found that the parameter is slightly smaller. I selected several more NEXTVAL, the parameter caught up with the data and eventually the error message disappeared. The DB then ran healthily for several more months before it was scrapped.

I don't know what it got out-of-sync with. I also don't know whether it was caused by the restoration. It's not easy to corrupt an oracle db and start it up.

As a supplementary, in Oracle, a sequence is only for generating unique values. It does not guarantee to be contiguous nor imply anything on sequence of events.

Re:PostgreSQL: Why don't people use it that much?

[Thundersnatch]

Or you could just use a GUID-generating algorithm that produces identifiers which are monotonically increasing (yes, it is possible) - MSSQL even has a special function for that, NEWSEQUENTIALID(), and of course the same thing can be done on client side as well.

Using a NEWSEQUENTIALID() sort of function, even on the client, creates a bottleneck: multiple threads must coordinate with some central location that stores the last updated value. Global shared state doesn't scale, and it makes code ugly. Using a per-server or per-client prefix value like a hashed MAC address as the part of the key also creates (to a lesser degree) the same problem, in that inserts do not always go to the end of a table.

Re:PostgreSQL: Why don't people use it that much?

[Splab]

Some accounting applications requires series to be ever incrementing with no holes, so yeah there are these requirements, they are however not very common.

PostgreSQL

[Anonymous Coward]

Just move to PostgreSQL. Jeez!

A forked turd still smells.

Haven't we be here already?

[Anonymous Coward]

The REAL MySQL is whichever has the trademarks. MariaDB will be another RMSDB. May even be better than MySQl, but it won't be MySQL.

One thing most FOSS people have to learn. In the corporate world, developers are resources, they are not what makes the product. The product is made by the Company, using whichever resources it has (cash, materials, people, etc).

Official is irrelevant

[davidwr]

If the unofficial one gets broader support, it will probably get a new name and MySQL will become irrelevant.

MySql

[inKubus]

Personally it's just more of the same from MySql in general. MySql AB didn't do much of anything with it since 5.0 came out. They wasted a lot of time on a complete rebuild, on adding more features no one cared about. The thing about MyQql 5.0 is that it's really not a very good database. MyISAM sucks (but it is small and fast..) and InnoDB is bloaty. So I think that really MariaDB is going to be the future. Of the codebase. That is the nice thing about MySql is that really it's a wrapper around the Storage Engines. But the problem is the wrapper sucks. No kerberos/LDAP authenication?! What?

I could see Oracle taking one of their open databases and adding a Mysql compatibility layer so basically you can run stuff designed for Mysql on Oracle. This is really their bread and butter already, they move legacy stuff off old UNIX and IBM databases into their DB. Look at all the gateways 9i had. MySql only implements a subset of what Oracle can do. And with no support for the more modern, more object orientated practices, along with trees, etc, I don't see MySql making it out of it's current place as a cheap small database for non-critical applications.

That's not to say you can't make it quite stable and fast but it's not that out of the box. And the fact that 5.1 shipped with a crashing bug really makes me doubt Sun's desire to continue the brand. Which brings me to the forks, which are really the only thing keeping a stable 5.1 version alive out there.

Postgres is really not a viable replacement because it's a database nerd's database. I like it, but the data analysts at work won't be able to deal with its quirks. It does do a lot, but not small and fast like MySql. It comes from a long line of great database researchers, all of whom are well known around the Valley. A lot of all the major players' databases in the valley are based on ideas from Ingres including Oracle.

Personally, I think SQLite3 (4) is going to be the database of choice for small web hosts very soon. Small, portable, fast enough. At that point MySql will no longer have a purpose unless they can move into the middle tier dominated by MS-SQL.

Re:MySql

[jd]

Surely you want a high-level authentication mechanism like SASL, and allow the mechanism to use underlying engines like Kerberos.

Re:MySql

[loufoque]

Personally, I think SQLite3 (4) is going to be the database of choice for small web hosts very soon. Small, portable, fast enough.

SQLite is fast all right, but it doesn't scale at all.

PostgreSQL scales well, but is fairly slow on average.

The thing with MySQL is that you were supposed to have both.

Re:MySql

[inKubus]

Sure, but Mysql doesn't have that. Here's a really old worklog for pluggable authentication. There's a blog on it also.

Re:MySql

[Anonymous Coward]

Personally, I think SQLite3 (4) is going to be the database of choice for small web hosts very soon. Small, portable, fast enough.

Although there's been a push against it, I wouldn't have objected if Hixie had spec'd sqlite for DOM storage on the client. SQLite is untyped which I think is a more sensible approach anyway even if it makes DBA's wince. "let the db do the work" they say, so we create indicies, triggers and views but programers cannot avoiding type checks and/or casting. These days I only use a few data types, storing dates as int, IP's (long form) as int and so on. Not only is code like that more portable, it saves me having to remember datatypes for whatever db I'm working with this week.

SQLite has issues with write concurrency and I'd like to see a simple, single user daemon using unix sockets (named pipes on win). In keeping with the lite design, access control for this userspace daemon should be provided by the file system not some master table. With something like that, most of us could probably avoid ever having to deal with heavyweight databases ever again.

Re:MySql

[Just Some Guy]

SQLite is fast all right, but it doesn't scale at all.

MySQL does?

PostgreSQL scales well, but is fairly slow on average.

Honestly, that hasn't been true in years.

The thing with MySQL is that you were supposed to have both.

And yet somehow ended up with neither.

Re:MySql

[shish]

PostgreSQL scales well, but is fairly slow on average.

I spent months making my queries ugly and hacky and optimised for mysql, and my website would still grind at about 500 concurrent users. Out of curiosity I switched to postgres, code unchanged other than the minimum to be compatible and it ran fine up to around 1000. Then I rewrote the queries in order to make them simple and elegant, and postgres ran them /even faster/ :P

Re:MySql

[Carnildo]

SQLite is fast all right, but it doesn't scale at all.

MySQL does?

In comparison? Yes. When you're doing simultaneous writes from multiple threads, SQLite's lack of fine-grained locking kills performance.

A couple of months back I did some benchmarking of MySQL and SQLite3 as possible storage backends for a website. The test load was a task consisting of two SELECTs, an UPDATE, and an INSERT. With only one or two simultaneous clients, SQLite3 was faster. With three clients, they were tied. At 50 clients, SQLite3 was taking over 100 times longer than MySQL.

Re:MySql

[mcrbids]

PostgreSQL scales well, but is fairly slow on average.

Really? Because any recent review of Postgres shows that it stomps the pants off MySQL in all cases except very simple queries against very small tables. (And really, who gives a !@#% about that scenario?)

Also....

1) It's data validation is excellent.

2) It's extremely stable. In YEARS of working with it, I've had ZERO integrity issues despite managing rather large data sets.

3) Particularly important: it maintains good performance as the query complexity soars. While it can take a bit of tuning, I've done 12 table joins with combined inner, outer, and subqueries and millions of records, with an average return time of around 0.2 seconds. The statement alone was two pages, printed form, on a single-core Athlon 64 with ATAPI drives and 1 GB of RAM.

4) It's FREE FREE FREE!

5) It includes excellent near-realtime replication. (functionally analogous to MySQL replication, which is nice when it works, but since it usually doesn't, well...)

1994 called. It wants its stale information back.

Re:MySql

[Anonymous Coward]

PostgreSQL scales well, but is fairly slow on average.

PostgreSQL had/has a reputation of being slow compared to MySQL mostly because the 7.x series default postgresql.conf file limited the it to using 16MB of memory. This advantage became negligible once you adjusted the config to something more reasonable.

The thing with MySQL is that you were supposed to have both.

MySQL was actually known not to scale initially.

At the time MySQL only had the MyISAM backend, and it's still the default one. The ISAM algorithm/data structures were known (for decades) to be fast, efficient, and easy to implement. However, they have severe read/write concurrency issues inherent in the design (same reason why the backend only supports table level locking). Also it lacks the ability to have major features (like transactions) that people had come to expect from a db. Thats why others didn't use them as a primary backend.

InnoDB was really what made MySQL a viable general purpose DB

SQLite is fast all right, but it doesn't scale at all.

MyISAM has concurrency issues at the table level since the table data was stored essentially in one flat file.

SQLite, while totally awesome for its purpose as an embedded database, stores the entire database (not just each table) in a flat file. Its concurrency issues are therefore understandably worse.

Re:MySql

[raddan]

But wait-- is it really that important for the database to do authentication anyway? It's not like every single user of your application needs different database rights. In the apps I write, there's usually like three user-types with differing permissions. The database maintains a map between the userid and the user type on the DB. It really doesn't need to know much more than that. As long as the user really is who they say they are-- and that part can be handled further up in the stack-- everything works simply.

Re:MySql

[Requiem18th]

(And really, who gives a !@#% about that scenario?)

Why sqlite of course! And they are unbeatable in that niche.

Re:MySql

[shentino]

That's kinda the same issue you get when you have a gazillion threads fighting over the same core.

Should databases schedule? Give each client a slice of time to have exclusive access to the database?

Just throwing this idea into the ring, I'll let the gurus dissect it mercilessly.

Re:MySql

[NoelProf]

The perception that MySQL is smaller and faster than PostgreSQL is bullshit. PostgreSQL is actually smaller than MySQL even though it has many more features. In my experience Postgres's query planner is light-years ahead of MySQL. The result is that PG is almost always faster than MySQL in the general case. MySQL is also loaded with gotcha's and odd behaviour based on the "back end" or what settings you use.

Re:MySql

[Cico71]

The result is that PG is almost always faster than MySQL in the general case.

Amen. And that is exactly why DBMSs where built in the first place: to handle efficiently, and independently from physical implementations, the the general case. A lot of solutions that implements very specific, hard-coded, physical access paths are faster than any DBMS.

MySQL is also loaded with gotcha's and odd behavior based on the "back end" or what settings you use.

And this is exactly what a DBMS should prevent: dependencies on physical implementations.

If someone think this is only theory, then he should do a reality check and see what good optimizers in the Oracle, DB2, MSSQL (and I guess PostgreSQL but I have not experience there) space can do with extremely complex queries

If I need to be extremely fast just a bunch classes of queries for a web back-end, then I may as well replicate data and just use an indexing engine.

Re:MySql

[Anonymous Coward]

I love SQLite. Super fast, lightweight, tiny.

But if you try to index a column that wont fit in memory and it has to go to disk, it will take longer than the lifetime of the universe to get the job done. I left my machine crunching on this task for a few days before I gave up and went to Postgres, which did the same task in seconds.

Re:MySql

[kalman5]

And why you didn't test postgresq as well ?

Re:Oracle needs to cater to business not the commu

[amicusNYCL]

Your company must be run by morons.

Obviously. In fact, the people who run Wikipedia, craigslist, youtube, Slashdot, Apple, Cisco, Cray, Dell, Intel, HP, Motorola, NEC, TI, Xerox, Adobe, Symantec, Novell, McAfee, Citrix, Continental, Orbitz, Priceline, Amazon, ebay, Google, iStockphoto, Pricegrabber, Yahoo, ZipRealty, Linden, Audiogalaxy, Digg, del.icio.us, Facebook, Flickr, Freshmeat, LinkedIn, Photobucket, Stumbleupon, Twitter, and WordPress are clearly morons for using MySQL. These people don't know anything about databases.

In fact, I'm pretty sure you're the only person who's not a moron.

How healthy are forks?

[rockmuelle]

Are forks really a fact of life for healthy open source projects?

Most open source tools I use on a regular basis have never been (successfully) forked. More importantly, none of the other members in the LAMP stack have undergone a forking of the scale that MySQL is about go through. Sure, there are multiple 'L' distributions, but there isn't a healthy eco-system of production forks at the kernel level.

-Chris

Re:How healthy are forks?

[jbolden]

Most open source tools I use on a regular basis have never been (successfully) forked.

I doubt that;

GCC
which is a compiler for C which is a fork of B
XWindows 2x
LaTeX (from TeX)
all the shells are forks
Perl is a fork of Sed/Awk
Python and Ruby are forms of Perl
Oracle itself is a fork System R
etc... etc

Re:How healthy are forks?

[Kompressor]

For one (admittedly extreme) example, GCC -> EGCS. The fork was so successful that it was eventually accepted by the FSF as the official GCC.

Re:How healthy are forks?

[RevDiaBLo]

I guess it depends how you define a fork. If you consider a patch set a fork, then most distributions fork a large majority of the software they ship. And for the kernel, there are actually many many forks--it's the development model they actively encourage with git.

Re:How healthy are forks?

[Phroggy]

Firefox (then called Phoenix) was a fork from the Mozilla Suite. That worked out pretty well.

Re:How healthy are forks?

[Grishnakh]

The two biggest success cases are XFree86->X.org, and GCC->egcs->GCC. Both of these were resounding successes.

XFree86 was widely criticized for moving too slow, not integrating enough new features, and having someone in charge that everyone hated. Finally, XFree86 instituted a license change that was incompatible with GPLed projects, so they forked it to X.org. All the best developers abandoned XFree86 and moved to X.org. Now, just about every Linux/BSD distro uses X.org, and no one pays attention to XFree86 any more. X.org has tons of updates, improvements, and additions which were being held back by the XFree86 leadership.

GCC has a similar story, though it's a few years older. gcc was getting decrepit, so a new renegade team started egcs. This incorporated new features, revamped the codebase, etc., and eventually everyone was using egcs instead of gcc. Finally, the gcc team just disbanded as they were no longer relevant, and egcs was renamed to gcc.

Of course, there's probably some cases of forks where the forked version went nowhere, and no one remembers it. But that doesn't matter, as it didn't hurt to have the fork (except the guy who started the fork, and ended up wasting his time on it, but that was the risk he took).

Forking generally works extremely well. The ability to fork keeps projects competitive and responsive, lest some pissed-off developers decide to fork it. It keeps maintainers from becoming dictators who only hinder the project's development, instead of fostering it, because it allows the developers to take control away from the maintainer when they've had too much.

It's too bad companies can't work more like this. There's plenty of companies where the company's owners or management are idiots and drive the company into the ground, and all that work wouldn't be lost if the employees could stage a coup, so to speak, and take over. Since software can be copied at zero cost, and OSS licenses allow forking, this can and does actually happen in the open-source software world.

Re:How healthy are forks?

[Grishnakh]

Sorry for replying to myself, but another fork in the making is OpenOffice.org->Go-OO. This one is interesting because it's more similar to the MySQL fork, in that the main project is controlled directly by a big corporation, rather than a nonprofit entity as GCC and XFree86 were. Apparently a lot of OO.o contributors weren't happy with the control Sun had over OO.o, and believed that Sun wasn't very accepting of patches contributed from outside Sun. So they forked it to "Go-OO", which is now included by default in several Linux distros.

Re:How healthy are forks?

[derGoldstein]

From my experience, it serves two purposes:

1) Internally, anyone who's working on the project and wants to fork it in order to make disruptive changes, can do so in a "bubble" so that it doesn't effect the main branch. Yeah, I know that many people don't really consider that "a Fork", but rather a "Branch". In practice, however, there's very little difference. It's just perceptional.

2) It's a penalty that can be 'exacted' in case the overseeing developers/organization (see this story) have become detrimental to the project.

I suppose you can look at the first type of forking as beneficial and healthy (even as a type of internal natural selection process), while looking at the second one as the punishment that would befall "wrongdoers to the project".

Off the top of my head, though, two successful forks: Webkit and Inkscape.

(and I'm sure there are other beneficial uses of forks, I was referring to the primary types -- and again, only from my experience)

Re:How healthy are forks?

[Grishnakh]

I don't this is quite accurate.

C isn't a fork of B, it's a successor. It was even made by the same people as C.

I don't know what "XWindows 2x" is.

LaTeX isn't a fork of TeX, it's an add-on. LaTeX uses TeX as its backend; LaTeX takes stuff written for it, translates it to the lower-level TeX code, and then runs the TeX interpreter on that.

The Unix shells copy a lot from each other in terms of features, but they're certainly not forks as they share no code.

Perl isn't a fork. It copies features from sed, awk, sh-shell, C, and other places. It shares no code with those.

etc.

You seem to not understand what a "fork" is. A fork is when someone takes the source code to a project, and starts making their own changes to it under their own control, effectively making a new project, but reusing all the old code so they don't have to start from scratch. Obviously, this only really happens in open-source projects since proprietary ones always stay in control of the owner.

Calling Perl a "fork" of sed/awk is like calling Peter Jackson's "Lord of the Rings" trilogy a "fork" of the old animated movie from the 80s by the same name.

Re:How healthy are forks?

[Grishnakh]

1) Internally, anyone who's working on the project and wants to fork it in order to make disruptive changes, can do so in a "bubble" so that it doesn't effect the main branch. Yeah, I know that many people don't really consider that "a Fork", but rather a "Branch". In practice, however, there's very little difference. It's just perceptional.

I think there's a big difference. A "branch" is when I, a developer at work, make a new branch of my project so I can test out new changes. I even check in these changes in our SCM system so that I don't lose them in case my HD dies. Eventually, after working on it for a while (perhaps in conjunction with other team members), we decide these changes should be merged into the trunk for full testing by QA. We do, and the project continues with these changes.

That's a branch, and is a normal and nearly indispensable part of software development.

A fork would be some spies coming to my office, hacking into my computer, and stealing our source code, and then making some changes to that and using it for their own purposes. In the open-source world, this is done without any illegal activity, as the license explicitly allows this.

The difference is the purpose, and who's in control, and who's doing the branching/forking. Technically, there's no real difference as it's just a bunch of bits being copied. But at a higher level, it's a huge difference. A developer in control of a project branches his code all the time, just to develop new features. He almost never "forks" the code, because that would be pointless; he usually intends to merge his changes back into his own codebase. A fork happens when someone not in control of the project takes the code to make a similar project (under his own control), which starts out with all the same source code as the original project, but then move in a different direction without merging the changes back into the original codebase.

If you make no distinction between these two words, then you might as well call it a "fork" every time you edit a file. This obviously makes no sense, which is why we reserve the use of the work "fork" in software development to those rare times when someone creates a new project using the source code from another.

Re:How healthy are forks?

[jbolden]

In terms of XWindows it is the windowing system on Unix. 2X means two times.

Well I define fork more broadly than that. In terms of functionality and user base not just code base.

TeX and Latex are a good example. At the time LaTeX "forked" from TeX there were differences in how the system would operate in terms of things like higher order structure. Lets of TeX commands aren't supported (at least not safely) and the basic structures changed. Yes the Web source code between TeX remained unchanged but there were two users communities and they products developed independently of one another for years features migrated from TeX to LateX.... A classic fork, a collaborative one. PDFTeX is also a fork in the same way.

As for your description of LaTeX it is not quite true. LaTeX is a version of TeX environment that is heavily modified via. macros. There is no "compiler" step that translates the LaTeX code into TeX code (unless you are including the entire LaTeX code base in the translation) because things like counters are introduced.

Re:How healthy are forks?

[Grishnakh]

I'm sorry, but you have no idea what you're talking about. Read this for more info. Here's an important excerpt:

LaTeX is intended to provide a high-level language that accesses the power of TeX. LaTeX essentially comprises a collection of TeX macros and a program to process LaTeX documents. Because the TeX formatting commands are very low-level, it is usually much simpler for end-users to use LaTeX.

However, pdfTeX is a fork.

LaTeX still requires a TeX engine, either original TeX, or pdfTeX, to work.

In terms of XWindows it is the windowing system on Unix. 2X means two times.

I'm well aware what the X Window System (not "Xwindows") is. What I don't know is what "Xwindows 2X" is supposed to be. A google search yields absolutely nothing.

Re:How healthy are forks?

[shutdown -p now]

And one more fork in the making that looks like it is going to have the same impact as XFree/X.org eventually, if not bigger: EGLIBC

Re:How healthy are forks?

[derGoldstein]

If this was an official/accepted definition, I would completely agree. In fact, if there was a standards committee that offered what you've written above as a proposal, I would back said proposal, if only so we could finally agree on the proper use of the term "branch".
This is not the case, however. Check out github and the terminology they use. If you use that site/service to manage your project, there's effectively no difference between a "fork" and a "branch". This is why I said this was perceptional, because really, it's like you said:
'you might as well call it a "fork" every time you edit a file'
In practice, on a mechanical level, it's exactly the same. This is why different people use "fork" in whatever context they choose.
I agree with your definition, but not everyone does.

Also, there are quite a few reasons why a proper "fork", not a "branch", would be beneficial. You could be leading a project, and one of the developers could decide to try a different approach, to a drastic extent. They would fork the codebase and start developing separately, to the point where there would be no way to merge his code with the main project. This new "subproject" could become a subset of the primary project, designed to deal with specific tasks.

At this point, the new code could remain under the same "roof", or that developer could opt to lead that fork independently. The difference between those two options would be great, but in the end they're both "forks", even by your definition (a new project is created, or "spawned").

In the end we're arguing semantics. I could choose to use your terminology, but I would have to preface what I mean whenever I do, so that people understand that when I say "fork", I specifically mean something that could not be merged back into the main project.

Re:How healthy are forks?

[Znork]

What I don't know is what "Xwindows 2X" is supposed to be.

I think he means that X has forked twice. Which is sort of correct if one considers 'X' to have been XFree86 and the original fork of X386 to be the evolutionary path of the X server in question, ie, X386->XFree86->X.org.

Of course, a more complete view of X as the entire X code base would indicate many more forks, as pretty much every proprietary vendor-specific fork would be one too. With partial re-basings and some flow both ways, but still forks. Well, unless one considers them just very independent branches. So it's debatable I guess, but at least the XFree86 chain seems to be widely considered to be forks.

Re:How healthy are forks?

[damaki]

More important than "several Linux distros", it's included in Ubuntu, the most hyped Linux distro.

The strange part is that the patches are actually affecting missing stuff that seriously hinders OpenOffice.org capabilities to be compatible with MS Office. But I know that Sun people are control freaks and do not use much external code. Maybe this could make them realize their policy is stupid and unsustainable.

Re:How healthy are forks?

[shentino]

I wouldn't let my employees coup and take over, in fact, I would can anyone who even suggested it, union protection be damned. I could easily fake it as a performance issue.

And since according to work-for-hire I own all the code anyway, any of my peons who don't like my decisions can either suck it up and kiss my ass, or go out into the cold and try to do it themselves from scratch, without getting my legal department on their behinds for "copyright infringement" or "breach of non-disclosure agreement". Being powerful really rocks, and I love being in charge, especially when I get to milk my workers for all their worth and then can them when they are of no use. Bonus points if I'm rich enough to keep my congress critters in my pocket keeping the laws off my back.

I am, of course, being sarcastic, but often there's a lot more to software development than simply doing a good job as a software engineer.

Re:How healthy are forks?

[TheTurtlesMoves]

..Sun people are control freaks...

Its more of what cooperations tend to be like. Even when everyone likes the idea of open development etc. You still have n+1 PHB and lawyers to keep happy. Thats why a corporation "sponsoring" a OS projects tends to work, while a corporation leading a OS project tends to gain this "control freaks" reputation. Of course they may really be control freaks too.

Re:How healthy are forks?

[bytesex]

"Well I define fork more broadly than that." Yeah and I can define 'two cows and a daisy' to mean 'man I had too much to drink last night', but that doesn't make it so.

Re:How healthy are forks?

[Grishnakh]

I am, of course, being sarcastic, but often there's a lot more to software development than simply doing a good job as a software engineer.

Like what?

I wasn't saying that employees should be able to stage a coup and take over the company; that's a basic part of property rights. If I own the company, I should be able to maintain ownership, no matter how bad a job I do, even if I run it into the ground. Employees taking over the company would deprive me of my property rights, since the company can't exactly be duplicated at no cost.

The nice thing about software is that this doesn't hold true. I can make a copy of the XFree86 source code easily, then make my own changes and release it as something better (with a different name of course), without depriving the original XFree86 guys of their copy. So nothing is lost.

As for "a lot more to software development", like what? Brilliant management? Sorry, I haven't been seeing much of that from corporations these days. All software development needs is good software engineers, and a good project leader to steer them in the right direction. This might not be the most profitable way of doing things, but the best software I've used is usually not the most profitable; exactly the opposite, in fact. The most profitable (for its maker) software I've used tends to be the worst. What good software needs is NOT a bunch of PHBs and useless marketing types to screw it up.

Re:How healthy are forks?

[jadavis]

XFree86->X.org, and GCC->egcs->GCC

Both of these examples started from something with very little competition: XFree86 and GCC were virtual monopolies (or as close as one can get in the free software world).

MySQL doesn't enjoy quite the same status. A lot of the popularity of MySQL is on the low end (I know there are big companies using it too -- I'm talking raw numbers), and those developers are quite finicky. They could jump to PostgreSQL, FirebirdSQL, SQLite, or a non-SQL DBMS. Also, on the high end, people will be reluctant to jump to a fork of uncertain future.

there's probably some cases of forks where the forked version went nowhere, and no one remembers it. But that doesn't matter, as it didn't hurt to have the fork

I'm not so sure that a fork is painless. MySQL is a platform, and depends on application developers to keep it interesting. A fork pulls application developers away, and in the process they could be lost to alternative DBMSs.

Much of the power of MySQL exists because it's the "default" DBMS. If a fork starts to become viable without completely dominating (and completely dominating is unlikely), the "default" choice then presents more choices that have no default.

There are a lot of people in a "wait and see" mode with MySQL. What will Oracle do to the mainline? Which of these forks will be viable? When will they be stable enough to run production systems? When will the application developer community decide which one is the new "standard" MySQL?

If another DBMS product has an exciting release, a large fraction of these people will say "I'll check that out while I'm waiting" and never come back.

Re:How healthy are forks?

[shentino]

If the company holds copyright over the code, then no you can't actually do that.

Your idea is wonderful in theory, but it currently ignores the practical reality of how much of a clusterfuck IP law is today.

Re:How healthy are forks?

[Grishnakh]

I'm not talking about closed-source code here; I'm pointing out how open-source code, licensed under an open-source license, allows things (like forking) that simply aren't possible in the proprietary world.

Oracle doesn't care

[Reality Master 201]

Oracle doesn't care about losing the trust of the MySQL community. They already have a database to sell; they're probably more interested in the vertical integration with OS, hardware, and programming language tools.

I wouldn't be surprised if they were happy to sell the MySQL business unit, or kill it completely.

Re:Oracle doesn't care

[anime_boy]

Oracle won't sell the business unit. Since the purchase of RDB and Rally from DEC in 1998, Oracle has not sold of a part of any acquisition.

FUBAR

[Locke2005]

The question is not "Has MySQL Forked Beyond Repair?", but rather "Has MySQL been Forked Up Beyond All Repair?"

Re:Oracle needs to cater to business not the commu

[CarpetShark]

If Oracle provides what business needs/wants .. and that's what they have been doing thus far. They will be fine. Nothing to see hear move along.

Really? I'm more tempted to think that whichever fork becomes the version in Debian (by being cooperative with the community, having the best features, whilst not going against free software policies) will be the one that gains momentum. It's hard to beat being available at the touch of a button in both a popular server-grade OS, and it's popular desktop-grade offshoot.

And Oracle should care...why?

[swordgeek]

I don't know this for sure, but I can't imagine that Oracle bought Sun for the sake of acquiring MySQL. In fact, the only thing they really might be interested in it for is to destroy it.

Sure, they could nurture it along, make it the entry-level to REAL (i.e. Oracle) databases, but that's expensive, uncertain, and seldom profitable in the long run. Crushing it in-house, encouraging division and fractionation in the community, and letting it become another piece of roadkill is easier and more effective.

HOWEVER, I doubt they'll bother. MySQL doesn't compete with Oracle, and it's not good enough (no, really--it's not) to foster from their commercial point of view. Realistically, they'll let it sit there, doing its thing, while they concentrate on what they bought Sun for: Vertical integration, bundled solutions, and a complementary server line and OS for their software.

Thinking that Oracle CARES much about MySQL is almost certainly a conceit.

Re:Oracle needs to cater to business not the commu

[moderatorrater]

Because Oracle licenses can cost millions in an economy where every dollar counts whereas MySQL is free and fairly reliable. When he says "unneeded", he's probably referring not only to features but to stability and all those things that make people consider MySQL a toy database. If you need multiple 9s reliability, use something else. If you just need things to be up 90%+ of the time without any of the fancy features that Oracle gives you, then MySQL is an easy choice. At that point it comes down to whether you want to use MySQL or a more robust free database, and when it hits that point MySQL has a strong advantage in ease of use.

I've heard the arguments that postgres is as easy as MySQL, and they're bullshit. MySQL has good GUI applications, a good command line application, market share that ensures a tutorial for everything under the sun, and an easy installer for windows and linux. Postgres has no good GUI applications that can compare with MySQL's, their command line application is just as good in its own way, and the market share that ensures you need to google multiple times to find the info you're looking for. Installing postgres is also a nightmare compared to MySQL.

To sum up: free > $millions, easy > full-featured (in many circumstances).

Re:Oracle needs to cater to business not the commu

[Anonymous Coward]

Amazon uses Oracle on Linux:
http://www.dba-oracle.com/oracle_news/2004_2_17_conway.htm

They do have some MySQL installations, but the big DBs are Oracle+Linux.

MySQL has features, and it is free, but it can't compare performance wise to a properly tuned Oracle installation.

Re:Oracle needs to cater to business not the commu

[dabooda]

haha that sounded like the 'things I care more about than you' rant by Dr Cox in Scrubs! ... oh wait, I'm the only one that watches that show

Re:Oracle needs to cater to business not the commu

[Anonymous Coward]

I'm pretty confident that most of those companies are running MS SQL, Oracle, and MySQL... among other things.

MySQL has it's place. It's place is most certainly not in mission critical, highly available, highly scalable solutions.
How many Enterprise Data Warehouses out there run on MySQL?

Also realize that many of the web-centric companies you listed there probably started out using MySQL because they couldn't afford anything else. Their architectures often have a high degree of redundancy and load distribution among commodotized hardware. This is certainly a niche where MySQL reins supreme.

Unfortunately, no matter how hard you wish upon a star (schema - DBA humor ha-ha!) MySQL is just not on the same level as MSSQL, Oracle, or even DB2. Sorry.

Don't believe me?
Go ahead and create a multi-TB database (on one server) with a few tables that are >100GB each. Do some performance benchmarking after scaling as best you can with each DB System and let me know how that works out for you.
Oh, did I mention that this has to be on one server that you replicate to the other side of the world with less than 30s of change difference between the two and that you also need local HA and global DR OUTSIDE of the replication you just setup?
Oh yeah; your data change rate is ~500MB /min.
Good luck!

Maybe copy & pasting things you read on the internet and calling other people names wasn't the best idea, huh?

Database abstraction layers people

[cenc]

The LAMP stack was always fundamentally flawed because the 'M' was not really an open source public project, where everything else was. We need to replace the M with a true open source project, and I hope the best project wins.

What everyone failed to do, that they should have done years ago, as a standard was build database abstraction layers. I know people will argue about performance bla, bla, bla but I don't buy that. We should have been doing it for years, exactly for this reason. Instead everyone was lazy, picked MYSQL because it was free/cheap/easy to learn.

Now, hopefully get off our lazy collective asses and build in database abstraction layers in to our web apps the way we should have from the start, or replace the 'M' with something really open source and public that does not belong to a company.

My only hope is that Oracle might do the right thing and cut it loose as a fully open source project to develop to its fully potential.

Re:Database abstraction layers people

[derGoldstein]

I propose we replace the 'M' with an 'R'.

Re:Database abstraction layers people

[ChunderDownunder]

Sorry, LARP is already taken.

Re:Database abstraction layers people

[Anonymous Coward]

I've had it, from now on I'm going to choose my web stack based purely on the acronym I get. Marketeers take note...

BLOC = BSD, Lighttpd, Oracle, C
LNSL = Linux, NGinx, SQLite, Lua
SCID = Solaris, Cherokee, Ingress, D
WAFT = Windows, Apache, Firebird, TCL
MMMM = Minix, Mathopd, MySQL, Modula-3
WISH = Windows, IIS, SQL Server, Haskell
BLIP = BSD, Lighttpd, Ingres, Pherlathon

Pherlathon is a new language I just envisioned, combining the most maligned aspects of PHP, Perl and Python. I think it'd be really popular.

Re:Database abstraction layers people

[Samah]

Personally I'd prefer "LLSK", being Linux, Lighttpd, SQLite, Kepler. Note: Kepler website seems to be down at the moment.

<3 Lua

Re:Database abstraction layers people

[raddan]

PHP already has a couple: PEAR::DB and AdoDB. And, of course, Perl has DBI.

I don't think the lack of abstraction is stopping anyone. We might call our stack LAMP, because people have a good idea what that means, even if our actual environment is OADP (OpenBSD, Apache, DBI, Perl). Same difference. We have LAMP machines, too, and they are pretty much equivalent.

Re:Database abstraction layers people

[LearnToSpell]

Usually what I go with now is:

Linux
Apache
Mod_perl
PostgreSQL

Re:Database abstraction layers people

[shutdown -p now]

Now, hopefully get off our lazy collective asses and build in database abstraction layers in to our web apps the way we should have from the start

It's much more tricky to make a proper DAL than many people think, especially when you go beyond the basic SELECT/INSERT/UPDATE. For example, different databases have different locking strategies (that difference is most striking between traditional locking, and MVCC systems), which means that code that might deadlock/fail on one RDBMS will happily execute on the other. If you happen to develop on the latter category only (e.g. you test on Oracle, Postgres and Firebird, but not MSSQL or DB2), then you wouldn't even be aware of the problem. And it is very hard to abstract away that sort of thing - the best you can do there is to treat the database as a much more dumb thing than it really is, and restrict your operations to the most basic ones. But then you're not using all the benefits modern RDBMS give you, and, really, with such an approach, you might just as well be using something primitive like MySQL.

Re:Database abstraction layers people

[Anonymous Coward]

This is a pointless post, but I've just got to say, I loled at WISH and Perlathon.

Re:Database abstraction layers people

[cenc]

When I think of DAL, I am thinking about apps such as a CMS that had it's own library of SQL optimized for a selection of popular databases. Yea, generic ones will likly work, but for most large web apps I would prefer the application have its own customized abstraction layer optimized for each database (a few choices would be nice, not everyone in existance).

The open source ones that come to mind are things like phpbb or perhaps joomla. So many of the php based open source projects give very few choices, because the sql is spread all over the place, especially in the mods, plugins, and whatever. Only recently have you started to see them at least put a leash on where they are located and standardize their use.

Re:Database abstraction layers people

[tiggertaebo]

I'm confused here - In what way is MySQL not a "true" open source project?

I do however agree with you that proper abstraction layers are a good practice and help reduce reliance on any particular DBMS.

Re:Database abstraction layers people

[LaminatorX]

I'll take L^2P^2 please (Linux Lighthttpd Postgres Python)

Re:Database abstraction layers people

[jadavis]

What everyone failed to do, that they should have done years ago, as a standard was build database abstraction layers. I know people will argue about performance

It has little to do with performance. The problem is the "abstraction" itself. What is the abstraction?

We like to pretend that everything is simple, and that all those features in SQL are just needless complexity. So, people start writing a new, simple model that "should work for 90% of people" (whatever that means is usually left unsaid).

Then, people actually start writing real business applications, and they run into problem areas that don't really fit the simple model any more. We would like to think these are edge cases, but in reality the simple model does not have a way to handle typical constraints which are derived directly from typical business rules. As a result, the data that comes out of the DBMS can't be trusted, and that results in either a completely wasted effort from the start, or mining your own data.

"I put it in a DBMS" is not any more meaningful than "I put it in a spreadsheet" or "I emailed it to you as a word document", unless that DBMS enforces useful constraints on your data that match your business rules.

SQL isn't perfect. Far from it. But it is a defined abstraction, it has an answer for a wide variety of business concerns, and it's supposed to be a standard. Energy is best spent in one of the following ways:
  * better support for the standard (which MySQL has largely rejected)
  * improve the standard
  * replace SQL with a language that's at least as powerful

Re:Oracle needs to cater to business not the commu

[renegadesx]

In the SMB space yet but dont forget Red Hat/CentOS still dominates the enterprise in the Linux space.

Who they go/stick with will be important too.

Re:Oracle needs to cater to business not the commu

[growse]

I've heard the arguments that postgres is as easy as MySQL, and they're bullshit.

Lets see:

Postgres has no good GUI applications that can compare with MySQL's

Why you need more than one is beyond me. Isn't Pgadmin enough?

their command line application is just as good in its own way

Well, ok. Whatever 'in its own way' means.

and the market share that ensures you need to google multiple times to find the info you're looking for.

Postgres has some of the best documentation of any open source project I've seen. Sure, MySQL is good as well, but lets not spread bullshit here.

Installing postgres is also a nightmare compared to MySQL.

You mean in a download-the-msi-and-double-click-on-it way, or the apt-get-install-it way?

To sum up: free > $millions, easy > full-featured (in many circumstances).

Well, it's fully-featured, but not necessarily all of those features at the same time. Try doing full-text indexing on a database with foreign-keys on it in MySQL sometime.

"Free" not "free"

[zuperduperman]

> MySQL is free and fairly reliable

Actually it's not free, it's "Free". That might seem a small distinction, but the minute you want to license your own code that talks to MySQL commerically it will bite you in the ass in a rather unpleasant way (as in, call your Oracle sales-rep).

Re:"Free" not "free"

[zyzko]

> MySQL is free and fairly reliable

Actually it's not free, it's "Free". That might seem a small distinction, but the minute you want to license your own code that talks to MySQL commerically it will bite you in the ass in a rather unpleasant way (as in, call your Oracle sales-rep).

The minute you want to sell closed-source derivate of MySQL it will bite you, just like Linux will. Commercial open source projects have no problems, and while MySQL sales rep will tell you otherwise legally redistributing your closed source program (with MySQL) which talks to unmodified MySQL via TCP or UNIX socket is quite safe. If you don't distribute MySQL at all you are perfectly safe. It's not exactly rocket science to provide instructions how to add the JDBC driver to the application server or install the needed DBI package.

But yes, PostgreSQL is "more free" in the sense that it is not GPL, and for some people it is important.

Re:Oracle needs to cater to business not the commu

[Achromatic1978]

Don't believe me? Go ahead and create a multi-TB database (on one server) with a few tables that are >100GB each. Do some performance benchmarking after scaling as best you can with each DB System and let me know how that works out for you. Oh, did I mention that this has to be on one server that you replicate to the other side of the world with less than 30s of change difference between the two and that you also need local HA and global DR OUTSIDE of the replication you just setup? Oh yeah; your data change rate is ~500MB /min. Good luck!

One server? Why? Your company is big enough that it can afford to be running OC-12 links at your site and OC-3 to both the remote, and the DR sites (500MB * 3 = 1.5GB/min, with TCP/IP overhead, in the order of 300mbps) for your database alone, let alone email and other connectivity, at going market rates of over $600k/mo, and yet you can't/don't want to cluster a couple of servers?

Re:Oracle needs to cater to business not the commu

[shish]

Wikipedia

Is 99% cached content, served as effectively static files

Google

Has some developers who have it installed; none of their major services are based on it

Facebook

According to the video of how it works, is basically a graph database, on top of a giant key:value table, on top of memcache, on top of mysql -- ie, they do their best to avoid actually hitting the database, and when they do, they only make use of the most basic functions (and it isn't even good enough for that, so they're working on a replacement)

Twitter

While large, it's hardly complicated in terms of database schema, and it /still/ falls over all the time :P

The other sites, I've never seen any behind the scenes reports on.

While "we have at least one copy of mysql installed somewhere in our organisation" is common, "we are using mysql as the central database for our heavily stressed mission critical systems, and it works great" is much rarer :-P

Re:Oracle needs to cater to business not the commu

[binary paladin]

MySQL's documentation is nowhere near as good as Postgres and when I found out this little tidbit about Postgres, I was sold:

http://www.postgresql.org/docs/8.3/static/datatype-character.html

In particular:

"Tip: There are no performance differences between these three types, apart from increased storage size when using the blank-padded type, and a few extra cycles to check the length when storing into a length-constrained column. While character(n) has performance advantages in some other database systems, it has no such advantages in PostgreSQL. In most situations text or character varying should be used instead."

There is more to learn when using Postgres, but not too much just to get up and running. I managed to get it running with an ORM that had been using MySQL in a few hours. The docs are really good too. Oh yeah, and my favorite animal is an elephant.

Where is the problem?

[drolli]

My impression is that mysql is in very widespread use. If beeing concerned about developer drain to forks, i am sure oracle will hire new developers. They may be not working as long on the project as the old ones, but i am sure oracle is not going to let mysql down. Oracle has a *big* advertisement over main entrance of the biggest public transport station in Tokyo, moreover they plaster nearly every airport i know with advertisements. My estimation is that all these advertisements today cost *much* more per effect they have today than the comparatively cheap development of mysql for influencing tomorrows decision makers. Moreover, selling support for mysql also gives them direct access to customers they otherwise would never get.

I own "oursql.org", anyone (legit) want it?

[oursqldude]

I had bought it for other reasons (see the site). But now it serves a new purpose.

Looks like the forks already have names though, I guess.

Re:I own "oursql.org", anyone (legit) want it?

[ModernGeek]

If you are really interested in getting rid of it for the obvious purposes of a fork, contact me. My email is pursini [ a t ] gmail [ d o t ] com

Re:Oracle needs to cater to business not the commu

[klubar]

With most of the exanmples mentioned, a small number of DB errors (lost records, index errors, sychronzation errors) are accepttable. For applications like youtube, slashdot, etc. it really doesn't matter if records are lost. I'd be more interested in hearing about major financial or business systems that run on MySql. MySQL is a nice database--good for hobby projects--even big ones--but nor really the choice for heavy lifting DBs. Most of those examples started with MySQL because they needed free more than reliable.

Re:Oracle needs to cater to business not the commu

[Hucko]

Not sure here you're right; if you assume the list to be accurate and true, there are a significant number of IT big boys in there...

Of course if you are right, then the list is most likely wrong.

Re:Oracle needs to cater to business not the commu

[Bill, Shooter of Bul]

Facebook issues reports everynow and then, but its a moving target as most of the sites are. Somethings only scale so far. Facebook & Google seem to be on the Map Reduce train for the majority of their services. Yet, Google is one of the major third party contributers to Mysql, although its not clear exactly where they use it ( I've heard rumors of adwords being done via mysql). They've reported uptimes of 2-6 years ( for the whole stack Hardware, OS & Mysql under load) here

Re:Oracle needs to cater to business not the commu

[Tolkien]

So I was only going to quote this bit:

MySQL has it's place. It's place is most certainly not in mission critical, highly available, highly scalable solutions. How many Enterprise Data Warehouses out there run on MySQL?

and respond, but then I decided to read the rest of your post first.

That said, this reply is thus a response to your post as a whole.

Are you seriously that stupid? If all those companies use it, it's for good reason. You're an idiot, Coward. I would mod you down if I could.

Re:Oracle needs to cater to business not the commu

[Anonymous Coward]

Because, to achieve superiority, one must alter context before presenting a challenge to ensure that, under these erroneous circumstances, the challenger can never win. Fair circumstances may have more accurate results, but they don't always please the uberleet.

Comment removed

[account_deleted]

Comment removed based on user account deletion

Irrelevant

[kriston]

It's irrelevant. We moved everything to PostgreSQL and life is as uncomplicated and standards-compliant as life can get.

Re:Oracle needs to cater to business not the commu

[Hooya]

I keep reading arguments about DBs that eventually end up claiming that - "well it's all well and good that youtube, facebook etc use MySQL but it's not really for the enterprise/heavy lifting/financial part so really, MySQL is no good."

Well then stop thinking of MySQL as an RDBMS if that pleases you. At least admit that it is good at something that other DBMSs aren't quite as good at - clearly there is a need for a 'less than ACID compliant DBMS' and MySQL fills that role quite well, thank you very much.

Re: Ingres

[butlerm]

Oracle and Ingres were serious competitors about 25 years ago. However, Oracle quickly adopted a significantly better design that put Ingres-like databases (Ingres, Informix, Illustra, the original Postgres, etc.) virtually out of business.

Not only that, the internal MVCC architecture of PostgreSQL is *much* closer to Oracle than any of the other Ingres derivatives - including Postgres itself. The original Ingres hit the wall in large part due to the lack of multiversion concurrency control and row level locking (among many other things). In other words, it was practically useless for running large transactions concurrent with other activity, an area where Oracle shines.

Page level locking and no MVCC means that writers block readers (and other writers) until the first transaction to lock the page commits. Page level locking means that which rows will be incidentally locked up by a transaction is essentially random.

Such databases worked well if you were read mostly, and ran small quick transactions during the day, and large long running transactions in appropriately scheduled batch jobs at night. Unfortunately, technologically speaking, MySQL is about to catch up to where Ingres was just prior to entering its terminal decline twenty years ago. The InnoDB engine is the only major exception. Of course people typically use MySQL for rather different applications than they use Oracle for, and MySQL is radically less expensive.

Comment removed

[account_deleted]

Comment removed based on user account deletion

Good Ole Oracle

[jandersen]

... Oracle will have to regain the the trust and support of the MySQL community -- in other words, 'stop acting like Oracle.

If you go to http://www.oracle.com/technology/index.html you may notice that they give away their base products (ie RDBMS and JDeveloper among others) to developers for free, and if you go to their forum: http://forums.oracle.com/forums/main.jspa?categoryID=84 you should find that they have a large and thriving community for that very reason.

I like MySQL - but Oracle is my favourite database by a mile. I do system work daily with all of the major ones: Oracle, DB2, Informix, Adabas, Sybase, Teradata. I'm no longer an expert in any of them (used to do a lot of Oracle), but I do know that Oracle is still the standard the others are trying to reach up to, SQL-wise and with respect to scripting.

MySQL is not even in that game IMO, but it has a very simple API, so it is hugely useful for smaller applications.

Mysql table level locks when ordering results

[Anonymous Coward]

We had awful performance nightmares when using Mysql5.0.x when ordering records on tables that had indexes. That created an enormous bottleneck as it was a table that was written to and read from often. We used InnoDB (garbage). Yes Mysql is faster, but only when using MyIsam (which is useless for me). After the pain of dealing with this, we migrated to Oracle10g. Yes we had teething issues (specially with storing 0-length strings!),but the app does fly now. Kudos to hibernate for making the application side fairly seamless. We had to change code where developers did not use binds, and hardcoded the values (e,g, boolean in mysql, and number in oracle). Very silly for us. Migration is a BIG pain, so you'll want to decide up front (from the business) how much REAL usage they expect so you can recommend the best DB for the job. Turns out business grossly underestimated the traffic and concurrent users.

Anyhow, I'm using Postgres 8.3 now and it seems rock solid, and no fudging around with IP permissions. Who in their right mind would run a database on a world accessible machine? Nuts! Oh and postgres doesn't appear to bottlneck on a sort.

Back to the original topic, stop making Mysql so hard to keep a track of. We started using Mysql 5.0.16 and ended up migrating all the way until 5.0.44 and it still was buggy until we gave up for the oracle solution.

Re:Mysql table level locks when ordering results

[daveime]

Who in their right mind would run a database on a world accessible machine

Someone who perhaps wanted to remotely administer the DB ?

The IP control on MySQL is so fine grained I have an account locked to my static IP only with separate login and password. I can restart it and do maintenance sat in my own comfy chair.

Hardly "world accessible", but never mind. You keep calling up your server host and paying $150 bucks service fee every time you want a restart of the MySQL daemon.

Bigger Question...

[Eskarel]

is does Oracle care?

It's not clear yet what Oracle plans for MySQL, whether they plan to use it to gain market share with companies where an Oracle solution is overkill, whether they're going to use it to create a cheap database appliance, or whether they plan to just scrap it.

Whatever their plans, having every loony in the bin fork the product and basically shatter the brand into a million pieces is probably the best thing they could hope for. Excessive forking, feature divergence and general uncertainty will kill MySQL for professional purposes which is all Oracle really cares about. If they want it dead it'll already be dead and if they want to use it, the "right" fork will be Oracle's.

Re:PostgreSQL FTMFW

[lamapper]

Sucks for me if I do NOT want to use Winblows platforms any more.

Re:Oracle needs to cater to business not the commu

[BrainInAJar]

but the big DBs are Oracle+Linux

Generally the Big DB's are Oracle/Solaris/SPARC...

Re:Oracle needs to cater to business not the commu

[zefrer]

"The downside is that they can't just merge the open code back into the commercial database, and that is a significant downside."

Says who? As far as I know Oracle has no plans to change MySQL's license.

Since both MySQL and its forks are open source there's nothing stopping them from merging the two. In fact that is what Maria DB is doing, merging changes to MySQL into it.

Re:Oracle needs to cater to business not the commu

[Zonnald]

That's weird,
Here http://www.postgresql.org/about/users Postgres claims Apple as a featured user.

Re:Oracle needs to cater to business not the commu

[Cico71]

I see calling names but no arguments from your part. "If they use it, there a reason" is hardly an argument anywhere except among clueless people.

Taking for granted that the list is accurate, I'm pretty sure none of them even remotely thinks about using MySQL for data warehousing. The ad-hoc nature of requests from a lot of business intelligence solutions, and the lack of optimizationn techniques in this scenarios, will simply bring it on it knees immediately. Let alone the fact that other products include specific storage engines for multi-dimensional models.

The other guy, OTOH, brought some arguments and gave due credit also to MySQL. So, sorry dude, but if I had one point left, instead of modding him up, I'd just mod you down into oblivion.

Re:Oracle needs to cater to business not the commu

[TheTurtlesMoves]

I have heard MySQL is better now. But when we moved a ansi sql thing to MySQL we where pretty surprised by what was missing. Back then the free version only had table locking, which made mostly read type database fast. But read/write/update heavy performance was quite poor. Also I hope transaction support and foreign keys are there now.

Postgres had all these features and basically a just worked with everything, including the store procedures.

But as i said that was a while ago now. Anyway we are currently all Oracle atm. But we have both mysql and postgres (and HSQLDB). A good database is one you can replace easily with another good database.

Open Source share isn't going anyware

[john_roy]

Oracle as a big issue here. The open-source share of the market is there an isn't going anywhere. Hopefully they realize that and fully support mysql, or they can sabotage the progect. Big companies start to realize that it's very difficult to gain open source market with proprietary products.
I believe for some applications there are room for both commercial(proprietary) and non commercial(free) software. I don't know if Databases are one of this applications, but Oracle seems to be doing pretty well.

On the other hand if they don't fully(and I mean fully) support mysql, mysql users will soon been migrating to other open source alternatives(and not oracle for sure).

Re:Oracle needs to cater to business not the commu

[zyzko]

With most of the exanmples mentioned, a small number of DB errors (lost records, index errors, sychronzation errors) are accepttable. For applications like youtube, slashdot, etc. it really doesn't matter if records are lost. I'd be more interested in hearing about major financial or business systems that run on MySql. MySQL is a nice database--good for hobby projects--even big ones--but nor really the choice for heavy lifting DBs. Most of those examples started with MySQL because they needed free more than reliable.

Actually I know several systems in the telecom, traffic and accounting industry where MySQL is used heavily and successfully. All these require outside oversight and reproductability of data (accounting, test result archiving, billing data). This is just a bullshit claim that MySQL is somehow "a hobby project database". I've seen a multi-hundred thousand dollar Oracle and MSSQL based systems have database errors, I have seen completely hosed PostgreSQL databases, I have repaired countless crashed MySQL indexes. There are means to counter these and if you just believe that your database magically works without logging, without redundancy and other things these "financial systems" need you are just waiting a disaster.

Right tool for the job and right precautions for disasters is invaluable skill to have, and no computer system I have worked with so far hasn't been failsafe without redundance and alternative ways of detecting what happened (outside the DB).

MySQL is actually exremely easy to repair with basic skills (MyISAM) and quite transaction-reliable (InnoDB), from these you have to choose which one to use. I'm sure PostgreSQL is also very good, but my personal experience has proven that it is not immortal either. And I don't expect my database ever to be, there must be a recovery plan for situtations like database corruption, error in program logic etc. when the integrity of data really matters. It's not just "but MySQL allows me to enter 2009-02-31 as a date" - in the real world this is not really very big concern (and for the record, you can now disable that behaviour).

HA HA HA

[Anonymous Coward]

Like Oracle gives a feces about community! They care about keeping the existing corporate supporting paying customers and then attracting more corporate support paying customers.

Re:Oracle needs to cater to business not the commu

[yelvington]

Parent was unfortunately moderated as "insightful" despite unsourced misinformation. Here's a sourced correction for you. Note that Google makes extensive use of MySQL in production:

Question: What's the biggest MySQL DB?

That's like asking what's the biggest Ferrari! What counts is performance and scalability. Omniture runs over 250 billion transactions per quarter on a farm of MySQL servers. Google uses MySQL for AdSense and AdWords. Other large installations include Wikipedia, Travelocity, Weather.com, etc. The databases can be hundreds of gigabytes. Sites run on hundreds of servers, some on thousands.

Read more: "How to Change the World: Ten Questions with Marten Mickos, CEO of MySQL" - http://blog.guykawasaki.com/2006/08/ten_questions_w_2.html#ixzz0GEf5dFmI&A

Open Source development is like cycling

[hayriye]

Open source development is like cycling. The main project is peloton. Normally, everybody goes with the peloton. But sometimes escape groups formed, which is a kind of fork. Some escapes are very short-lived, some continues for a while and swallowed by the peloton through the end of the stage, and some succeeds by getting a stage win.

Sometimes, if you are too slow, you can't keep pace with the peloton and keep touch with them. You never have a chance to catch them alone and abandon the race.

Re:Oracle needs to cater to business not the commu

[ThePhilips]

Go ahead and create a multi-TB database (on one server) with a few tables that are >100GB each. Do some performance benchmarking after scaling as best you can with each DB System and let me know how that works out for you.

Thanks for the great example!!! Now I know how 99.999% of DBs out there are!!!!!!

/sarcasm

I think so.

[geminidomino]

Yes, MySQL is well and truly forked...

Re:Oracle needs to cater to business not the commu

[geminidomino]

Installing postgres is also a nightmare compared to MySQL.

I think he means the "OMG It doesn't create a default superuser on install! Panic Panic!!!" sort of way...

Re:Oracle needs to cater to business not the commu

[jonaskoelker]

The do not need to provide the wining fork, just support it.

Wining, Diming, Forking... sounds like a good date(1).

"Open source" vs "Bazaar-style"?

[jonaskoelker]

We need to replace the M[ySQL] with a true open source project [...]

As far as I know,

• MySQL is available under the GPL (among other licenses).
• The GPL is an OSI-approved Open Source license.
• A project is "open source" if it's available under at least one OSI-approved Open Source license.

replace the 'M' with something really open source and public that does not belong to a company.

The FSF holds the copyright of many important pieces of open source software (though they prefer to emphasize that it's Free Software). They're not commercial in nature.

They're a non-commercial organization. Companies are commercial organization. The only difference I can see is the commercial aspect, yet commercial use and copying of both Open Source software and Free Software is allowed in the definition.

I think you wanted to address something other than the licensing. Maybe the development model? Copyright transfer? Governance and leadership?

They do tend to go hand-in-hand. But it's useful do distinguish different-yet-correlated concepts, such that we can discuss them more clearly and intelligently.

[That's also why RMS is against using the term "intellectual property".]

Re:"Open source" vs "Bazaar-style"?

[cenc]

yea, I guess what I am getting at is not so much the license as the development model / lack of community driven development. I would say however that the Mysql license is a bit fuzzy at best.

I do believe that commercial involvement and extension is often a good thing in open source, if not vital to success. It is more a matter of health of a technology. Both the current discussion of the future of Mysql and related debates about Java are symptoms of the insecurity among the users regarding the future proofing of important technology such as these.

It is something that has grown to be such a fundamental element on the web, that really does need more of a full community / public control. Hopefully, one of the forks will do that.

I believe more of an Apache (or similar) like community and partnership would be a good thing for Mysql where we can all sleep at night having a clear idea of where it is going, and more importantly that it is not going away.

What happens if say Oracle just pulls the plug (not likly)?

I have over 100 databases (not even sure how many) currently running Mysql. It would be economically devastating for my buisness, and I am small fry. I can adapt relatively easily, but it would still be an investment of time and money to do so.

Re:Oracle needs to cater to business not the commu

[amicusNYCL]

I'm not sure if you were responding to my post or the one yours is attached to (it doesn't list companies), but I just got those names off the MySQL site under "Customers". I was surprised to see some of those big names there (Cisco, Cray, Motorola, etc), and I definitely acknowledge that many of them can use MySQL for some purposes but not mission-critical stuff. It just irks me to see people trolling about MySQL being useless, it's obviously a useful piece of software that has a place in many applications.

Re:Oracle needs to cater to business not the commu

[Anonymous Coward]

Do you even know what a cluster is? Did you read the clip your quoted that says you "need local HA"?
Having a cluster != having the load spread out between multiple servers.

What makes anyone think

[teflaime]

that Oracle (Larry Ellison, really) WANTS to be the "right" side of the MySQL fork wars? It's fully possible that Oracle bought Sun to eliminate the current largest source of MySQL contract support, isn't it? I mean, let's face it...that's exactly the sort of think Ellison is known for (see PeopleSoft).

Re:Oracle needs to cater to business not the commu

[klubar]

I completely agree--MySQL is a great database for a lot of projects. If you have a relatively small (the definition of which can easily include > 10 million records) database, transaction logging is not a requirement, absolute concurrance is not needed, an occasionaly lost update or record is ok and off-site redundancy and failover are not needed it's a great database. For 90% of the databases, I'd heartily recommend MySQL (especially if developing in PHP). However for 90% of the database records, I'd recommend against MySQL and go with something that can meet the above requirements.

Truth be told, most of the databases don't need the heavy-weight overhead (CUP and money) of the commercial offerings, but no one should claim that there isn't any difference in quality, features and support between DB2, Oracle, MS SQL and MySQL.

Comment removed

[account_deleted]

Comment removed based on user account deletion

Re:Oracle needs to cater to business not the commu

[Anonymous Coward]

The other sites, I've never seen any behind the scenes reports on.

You obviously don't follow much about craigslist or yahoo finance, or Jeremy Zawodny's blog. Both use MySQL extremely heavily (albeit with various backend engines, not necessarily MyISAM).