How many trillions of dollars are spent on the occult, spirualism and religion every year? How many people die or are tormented due to religious conflicts and dogma?
And since if there is no free will then:
- rewards and punishments in an afterlife make no sense - since everyone's decisions are determined by their environment and their genetic make-up. Why punish or reward what is basically an extremely complex automation?
- criminals and other offenders deserve a little more understanding and efforts at rehabilitation than if they knew in advance that they were going to perform an 'evil' act and deliberately did so anyway.
So, if there is no free will, then most religous concepts of good/evil, guilt, responsbility and an afterlife are clearly wrong.
Maybe there's four products? If you also include:
- MySQL Cluster - a cluster (like Oracle RAC). Very fast, very cool. Except...all the data is kept in memory - so it is extremely limited in application. Also a separate codebase - so don't expect your applications to easily move to this platform. Also as far as speed goes, you can probably achieve most of the same results by just putting 12-24 gbytes or whatever of memory on your non-clustered database - and making sure the database uses it.
- Non-Free - the MySQL license is very complex. Many fans celebrate how the vendor is vibrant and healthy due to its revenue stream, but that the product is free. In reality the product is only free in very limited circumstances - and I often see it being used in situations in which the vendor should be paid. Of course, you need to consult with an appropriate lawyer if you really want to know whether or not you need to pay...
Sure, if these kids are going to become professional programmers. But if this class is intended to just give them a taste of programming or some simple skills - they discipline isn't the most important thing to learn.
Actually accomplishing something is important. And doing it with a minimum of distraction and fuss is also probably important. And if this is the case, then python is probably a nearly perfect first language. And since python tends to push you towards a right way of doing things, you will tend to pick up some valuable discplines there anyway.
I suppose you could also argue that VB might also apply since it's all about just getting things done, but it's so ugly that I think it'll tend to wreck the students for any other languages later on.
You know, the first language I learned way back when was COBOL. I didn't love it much, but that's only because nobody ever told me that you could create a Terminator with it!
Of course, even as a junior programmer I probably would have been sharp enough to send information directly to the brain on the cyborg rather than just doing a printout to the eye. But you know how it goes - machine generated code is always crap.
> The old one -- even the Democratic members -- did not.
Note that much of the support for the erosion of our personal liberties occured while this country was in a patriotic & fearful frenzy. Anyone in congress that took a moment to say, "um, could we talk about this for a minute?" was attacked by:
- president bush & his administration
- senate & house leaders
- media pundits
- etc
- about 50 million americans
- right wing talk radio & tv (fox news) All of which was enabled by most mainstream news media. Everyone was so interested in "working together" that they allowed the administration and their colleagues to manipulate everyone to pass this crap.
So, yeah many Democrats supported these intrusive policies, bills, etc. But often only because to attempt a sane action would have been the equivilent of spitting in the wind - and killed their chances to survive the next election - and eventually reverse this crap.
Remember the first invasion of iraq? When a large number of democrats pushed back? The Republicans often remind Democrats of how almost every one of those Democrats that pushed to avoid war lost their next elections. So, this time around the Democrats played it safe. Unfotunately, as much as I wish it wasn't true, that was probably the best policy for a minority party without house, senate or administration.
Of course, they could also have been doing it because they wanted government control or somehow benefitted from supporting it. But I think these latter reasons were the lesser.
It's interesting that some people think that the military spends efficiently but will then criticize the rest of government for being inefficient.
The reality is that they're all just massive bureacracies that waste money like mad - because everything takes too many review steps, too many approval steps and effective requirements gathering doesn't work this way.
> throwing out our perfectly good 2 year old body armor that we spent billions on.
well, there ya go - if you just spent billions on 2 year old, perfectly good body armor and find yourself having to throw it out - somebody made a mistake, huh?
I've been in the military, have seen a lot of money wasted - primarily due to process problems. For example, a small box of monitor 'wipes' was ordered (this was 20+ years ago). What appeared was a 700+ lb massive crate of thousands of individually-wrapped wipes. Total cost was something like $25k. These were just used as paper towels since it would have been too difficult to send it back.
> When talking about risks and countermeasures, it pays to be specific.
So, when talking to a bunch of slashdotters, the extra detail implied by the acronym 'IED' vs 'booby-trap' (or even just 'trap') has some kind of pay-off?
Like, some slashdotter out there may die trying to detect trip wires on a non-IED bobby-trap in his back yard?
Seriously, unnecessary precision in your speach can cause more confusion than it resolves.
> Remember how this started out - a sexual harassment lawsuit. I forget the details
You unfortunately then forgot about how the entire sexual harassment suit was funded and backed by an extreme far-right interest group. Not a group that historically has been concerned about women's rights (far from it), a group that sought to use any possible excuse to harass Clinton.
And sure, having sex with interns isn't good judgement, and everyone should be concerned with him about that. But frankly, I'm more concerned about one party disabling the government while creating a circus over a primarily private sexual relationship than I am about the president lying about who he has sex with. Lying about sex has no impact on the government, distracting the government and blowing millions of dollars to attack the president over a non-issue clearly does.
> Oracle is the shiznit when it comes to high performance general database work. It will scale far beyond almost everything else, > with DB2 a close #2. Niche players like TeraData have their place too, but only Oracle can scale across the entire enterprise.
Sure, if you're talking transactional systems (like airline reservations). But if you're talking about data warehousing, very large scale analytics - then db2 and teradata have the upper hand. Oracle's clustering is for failover, db2's clustering is for performance.
Many years ago I was a programmer in the marine corps - and our office was in the top of an aircraft hanger. The interruptions could include a sergeant that wanted you to empty waste baskets, carry some boxes around - or an A4, F4 or maybe c-130 racing up its engines directly outside of our door.
So, ear-protection was standard issue - and we simply got accustomed to working at our desks with this equipment on. And we got work done.
Years later, I've worked in all of the environments described here, and I prefer open-plan with a small number of coworkers (7 or less). A common understanding of the "rules" is required (no music, no hollaring, etc, etc) as well as some maturity from the participants. But this beats the pants off offices in my book - in which you deliberately sever communications between team members.
This is especially true if at the same time you're wondering how you're going to implement methodologies such as pair-programming: with two people per office you typically don't invite a third person in your office for daily programming sessions. So, where exactly would this occur? In some additional conference rooms? Unused offices? Well, assuming an infinite budget for real estate I suppose this is possible - but in the real world there is seldom so much vacant space.
> Um what do you mean about the lack of command-line features? SQL Server has only one interface and that interface is SQL text sent to it from a client.
Sorry, I should have been more clear. I was referring to the limited batch processing ability of windows: that is, windows/2000 doesn't have a reasonable shell built-in. So, to automate tasks you're stuck with the windows task scheduler, primitive bat scripts, etc.
On most sql server projects I ended up installing cygwin, and running batch processing via shell scripts and python. Eventually we changed some very large applications over from using DTS to using BCP and python - with a huge and immediate reduction in maintenance effort & failures.
But most sql server dbas that encountered these systems were shocked - this just wasn't "the way" to run sql server, and these dbas just wanted to use the gui - and were obstacles to really getting tasks automated. However, the seasoned oracle, db2, etc dbas immediately recognized what we were doing and had zero problems in transitioning over. And of course when we later converted from sql server to postgresql most of the platform-independent code survived just fine.
> 1. Can be true - but depends on how small the "small" projects are, and how many there are. 15 small Oracle databases are going to be more work than 15 Postgres > databases, even if you are already running Oracle for other stuff. Also, if there's enough work for the smaller stuff that it needs separate DBAs (even if > it's just 2 total), then getting Postgres ones would not be more expensive than Oracle ones.
I would only go with postgresql is it really was a separate team supporting them. Otherwise the oracle dbas are now going to have to also learn how to recover postgresql databases from backup, etc, etc. That just isn't worth it most of the time.
> 2. Blatantly untrue. Oracle will scale better on multiple machines, but on a single machine Postgres will outperform Oracle far more often than not.
Not for reporting. If I've got a four-way oracle server doing reporting with 100 million rows in its fact table partitioned on day with 2-3 million rows per day. Then oracle will be far faster at most trending (reporting) queries. You won't have 365+ tables to manage for fake paritioning, you'll get 4x the performance of the single threaded open source servers, and better memory management. It'll be *far* faster than the open source options.
> And I'd like to see some source showing Oracle's optimizer to be better than Postgres'. Well, I was really thinking about mysql here - which has historically had difficulty joining more than three tables without unpredictable performance results.
So, if you've got an internet-accessible search engine running sql server (wasn't my decision) then you can easily blow over $80k to license a four-way server. In fact the original estimates we got were over $100k.
This is a pretty trivial article which seems driven by ingres.
Anyhow, a few things that I'd consider:
1. since as the author mentions the open source databases aren't ideal for mission critical applications (yet), then many organizations will find themselves supporting multiple databases. Say, oracle for financials & crm & the corporate warehouse and postgresql for a variety of smaller projects. Makes sense in many ways - except: oracle is already free for the small databases anyway, and now you need the dbas to support multiple products. This is going to increase your labor costs - not decrease it.
2. for many large analytical databases (data warehouses, etc) the cost of using open source are actually higher than closed source. This is because db2, oracle, etc are better at using the hardware than the open source alternatives. They've got better optimizers, parallelism, far better partitioning, better better pool management, automatic query rewrite, etc. So, a $100k oracle lisense running on a $100-200k 4-way (i know, assumes discount) will out-perform postgresql (free) on a 16-way ($1m) in many ways.
3. for some applications mysql could be more expensive than oracle. Ok, not just because you need to do far more testing with mysql to make sure that none of the wacky silent errors are affecting your code. But also because of the odd licensing - that requires its own faq and tips to just license the product if you can't figure it out. Then there's enterprise db - not very familiar with this one, but I doubt that it is free. Meanwhile, at the low-end the big-three database vendors all support free products. So, whether or not you pay more may very well depend on how you use the software.
Of course, if you're at a company like mine, and get to bypass purchasing and just review the license & install - you probably are saving a vast amount of money after all.
I actually found sql server to be quite expensive - from licensing (which was running > $80k for a 4-way on enterprise edition) to labor.
The lack of command-line features meant that many operational activities that could be automated required a dba to manually do the job via the gui. And lets not even talk about how you had to completely recreate DTS packages when promoting them from dev to test to prod...
So, there are labor savings that you can get on sql server vs oracle, db2, postgresql, etc - but the lack of a command line interface wasn't a driver in my experience.
> When a computer doing the counting, you'd expect 100% accuracy.
no, there are legitimate reasons for inaccuracy, such as:
- user selects wrong candidate (perhaps they don't line up right in the display, etc)
- user selects a candidate, then backs up and selects a different one, but does this process incorrectly
- user's vote isn't completed and so entire ballot is discarded
- date is incorrect on voting machine
- etc
These aren't hardware or software errors, they're mostly human errors or usability defects. But historically I think this is where the problems are most likely to occur and have occured with paper ballots as well.
Of course, there could be something nefarious going on. I wouldn't instantly rule that out. Just wouldn't assume that just because someone thinks that there is zero chance of error here.
>> Many precincts are too small for generators to be practical, and UPS units also have a failure rate. What if, even >> though it was tested the week before, the generator fails on the day of the election? There is also the cost associated. >> Who is gonna pay for it all.
> Wouldn't that be like... fighting for the democracy ? > Sorry, I'm not an american, but I though you people didn't mind spending money while fighting for democracy. But maybe > I misunderstood, and all that money is for fighting for something else.
Sorry, but you're being naive. There's no way to ensure that a meteor or tornado doesn't strike, that a fire doesn't burn the building down, that the building doesn't plunge into a sinkhole, or that wild dogs don't kill everybody in the building and eat the computer.
Not likely scenarios. Neither is it likely that a well-designed computer system will fail. But they do occasionally. Hard drives fail, power supplies fail, memory fails, etc, etc, etc. Even if you go with redundant components the motherboard may crack, etc. If your objective is to create a voting machine that never fails you'll just spend billions of dollars and still only get 99.99999% availability.
So, be practical. Ensure that the system has good uptime (99.99%?), is extraordinarily difficult to hack, leaves a good audit trail, and that voting and auditing are an atomic action. That should be fine. And should be more valid than paper ballots.
LiquidCooled wrote: > I personally think mysql handles slash really nicely and coupled with the open source foundations it stands on allows > a lot more people to get systems up and running, moving to a proprietary database (even a free one) would exclude a lot of > people from running their own site with slashcode.
Yep - open source products are definitely more agile than closed source - especially if you have to work through a horrible centralized procurement process like I have to. However, I wouldn't assume that mysql helps slashdot out today. Aside from today's snafu:
- why has the addition of new functionality taken so long?
- why is it so painful to see all threads by subject over the last x days, filtering by y?
- why can't we easily see better metrics on threads?
- why can't we easily see better metrics on posters?
- etc Well, one reason is that the database would have to return aggregate results for many queries - and mysql is a bad match for mild analytical queries - in which you need to scan 10k rows and return fifty.
So yeah, mysql was about the best option five years ago, but has had commercial options almost as inexpensive for the past two years and open source options with better data quality and just as good performance for the past three years or so.
> Maybe there are enough database developers out there today who have been effected by this who could step up to the plate > and fix mysql itself so it does have the required features. There are certainly enough geeks in here.
Well, mysql has been getting much better over the years: just a few years ago:
- it had no transactions (and their vps said this was a good thing)
- no views
- no triggers
- no stored procedures
- no subselects
- a trivial optimizer (5-way joins were almost unrunnable)
- silent errors on data truncations (truncates numbers)
- silent errors on data types (allows feb 58th, 2004)
- silent errors on transaction-safe storage creation it's far better today, but still has inexcusable problems - in which a simple column addition requires an entire table to be rebuilt, adding an index to a 16m row table takes 3+ hours (!?), you can turn on error reporting for type failures, etc - but any client can override the server setting, etc.
So, they've come a long way, but are still only halfway there. The question for developers that want to jump in and help fix things - is where is their time better spent? Helping a for-profit company fix its product, or just using a completely unencumbered and free product like postgresql in which these kind of errors have never existed.
> I guess this is another thing to add to the MySQL gotchas page [sql-info.de]. Of course, in a decent database engine, > like PostgreSQL, if you alter a column data type then the indexes are updated to reflect this.
Nah, this is a feature - each storage engine for mysql can do it differently. Sometimes they might behave like any other database, at other times you might have extra steps. And you can even get this benefit within a single database across multiple tables.
FYI, regarding partitioning as a design consideration:
- Yes, it is well-worth looking at, but note that partitioning isn't as fast as btree indexes for retrieval - *if* you were only planning to grab a few rows. It's more of a reporting thing that helps save you from table scans when you need to grab thousands of rows.
- Also, I wouldn't necessarily trust the merge table or partitioning features out of mysql until others have worked out the kinks first: there's a lot of potential complexity there with how indexes work (global or local), how the optimizer works, etc.
- And you probably will have sql changes and maybe column changes - in order to incorporate something like a story-id, thread-id or time-id to partition upon and to use as a query predicate.
Regarding your product selection - if you're going to look at changes for partitioning anyway:
- it wouldn't hurt looking at a database with fewer sharp edges to injure yourself on either. I mean, really - you've got degraded functionality right now because of design defects in the tool
- which allowed the error to occur and take so long to reindex and process defects in that your process doesn't require the additional testing that is essential when using dangerous tools like mysql.
- side note: if you were using a database engine capable of reindexing quickly in parallel you would have had that index built in probably a fraction of the time. Sometimes it's worth spending a few dollars to get infrastructure that you can really depend on. Doesn't have to be much cash, I think the free version of db2 (limited to 2 cpus & 4 gbytes of memory) would probably be a substantial improvement.
1. ETL (extract/transform/load) processing in which source data is translated to the warehouse format. This consists of several components:
a. python-based etl framework - responsible for managing our processing queues. This code takes a decentralized approach quite different than most commercial etl tools - that are large and awkward monolithic applications. This has worked very well.
b. python-based transformation - typically simple custom-written programs that read a single file and write a single transformed file. Execution of these programs is handled by the etl framework - based on existance of data in the queue. Lookups are via python hash tables. This has proved to be an extremely easy to develop and maintain solution.
c. python-based, metadata-driven validator and transformer - warehouses are easy to load quite a lot of data quickly into (hitting easily over 100k rows / second) but very slow to delete data from. Additionally, many of our source systems will change business rules without informing us. So, we have a completely separate python validator that is handed each field individually, then writes the output row. It assumes that all business rules are handled by the calling transformation program - so it only double-checks type, date format, value ranges, default values. It also is ultimately responsible for final file formatting - this is because due to so many control characters in my data I need to go with fixed-length files, and formatting so many fields this way was so error-prone.
2. testing framework - we've got a few testing challenges in the warehouse:
- server configuration consistency between multiple read-only servers
- data consistency between sources and targets and between multiple read-only servers
- data quality due to lack of enforced foreign key & check constraints
- configuration policy compliance (corporate security and departmental standards)
- performance of complex queries Our tester generates tests for many of the above, and allows us to easily run tests for all of the above. We use it in both development for QA and then daily for continual QC (emails us alerts if there are any).
Really? hmmm, my team runs a huge data warehouse in which all data is transformed via python programs. This means 40+ million rows a day get hash table lookups, business rule validations, type changes and various transformations.
Then most of that data goes through an additional metadata-driven validation process that checks and confirms every single field. Again, metadata-driven - so a oouple hundred lines of various hash table lookups to determine exactly what validation applied to each field.
Ok, now let's talk performance:
1. io performance (most critical factor here)
a. same as c, in other words, extremely fast
2. cpu performance
a. much slower than c, but workable
b. psycho helps a ton
c. the ability to rewrite key functions and classes is cool
Let's also talk about productivity
1. great for ease in training part-time or junior programmers
2. great for immediately understanding old code
3. great for writing robust test-driven code
4. good for very quickly assembling a solution
The above combo of ease of writing, ease of maintaining puts it far ahead of perl, tcl and java. Ruby is really the one other one in this niche I can think of off the top of my head. And even there Python appears stronger on the maintenance side.
So, yeah - I'm sure there are many applications that have more significant performance requirements than ours does. But for this large application it is completely fine.
> Respected by those who agree with them, yes. Vilified by those who don't. The "Journalists" of Foxed News are probably even > more respected by more people. Why not then force taxpayers to subsidize Fox News as well?
Even most conservatives I know listen to NPR. I'm sure there are some of the Rush Limbaugh type who can't stand it. But keep in mind as a liberal I think it often reflects a much more pro-administration and conservative viewpoint than I do. Those that vilify it might be angered by the fact that they do give airtime to different views. That's got to really burn them up.
> If the forced money is so small, then they can do without it, right? Sure, punish them for being affordable while letting the military blow billions on pork-barel. That would be a great deal.
> The majority of which is CLASSICAL music. Which was creative and innovative...a couple of hundred years ago.
ah, you probably don't listen to public radio. Note that NPR != public radio, it is a set of services and syndicated programs oftened subscribed to by public radio stations. In my town I've got at least two public radio stations - one just plays classical music and is completely unaffiliated with NPR. The other is affiliated with NPR but plays no classical music. It does play local music, blues, jazz, worldbeat, bluegrass, celtic, zydeco and rock. Unlike a mainstream station it has a playlist as long as whatever the dj wants to play - instead of just 10-20 tunes.
Of course, your mileage may vary - but I've found pretty high quality of programming at all public radio stations that I've encountered so far.
On the other hand, if all you want is an extreme conservative point of view and top 10 country hits, then public radio probably isn't for you.
> If NPR wants to beg for money from listeners, it should have to compete in the open market. I shouldn't be forced to support them, at any amount.
Hmm, I'd think you would want all projects to work like public radio stations: they merely get seed money from the government, then have to augment that with voluntary contributions from listeners. That ends up taking around two weeks a year for my local station.
So, what you get from public radio:
- some of the most respected and professional journalism available in the US
- some of the most creative music across all genres available in the US
- high quality programming (Praerie Home Companion, This American Life, Wait Wait Don't Tell Me, etc)
- local news And what does it cost you? probably a few pennies a year in gov taxes plus whatever you care to donate.
Now, imagine if the military or some other truly huge government activities were funded this way! Don't like our boondogle in Iraq? Great, not only can you hopefully vote the bums out, but you can also withhold revenue. Having spent years in the military I can tell you that it is much less professionally run than my local *public* radio station!
Slashdot Mirror
...then it does matter
How many trillions of dollars are spent on the occult, spirualism and religion every year? How many people die or are tormented due to religious conflicts and dogma?
And since if there is no free will then:
- rewards and punishments in an afterlife make no sense - since everyone's decisions are determined by their environment and their genetic make-up. Why punish or reward what is basically an extremely complex automation?
- criminals and other offenders deserve a little more understanding and efforts at rehabilitation than if they knew in advance that they were going to perform an 'evil' act and deliberately did so anyway.
So, if there is no free will, then most religous concepts of good/evil, guilt, responsbility and an afterlife are clearly wrong.
Maybe there's four products? If you also include:
- MySQL Cluster - a cluster (like Oracle RAC). Very fast, very cool. Except...all the data is kept in memory - so it is extremely limited in application. Also a separate codebase - so don't expect your applications to easily move to this platform. Also as far as speed goes, you can probably achieve most of the same results by just putting 12-24 gbytes or whatever of memory on your non-clustered database - and making sure the database uses it.
- Non-Free - the MySQL license is very complex. Many fans celebrate how the vendor is vibrant and healthy due to its revenue stream, but that the product is free. In reality the product is only free in very limited circumstances - and I often see it being used in situations in which the vendor should be paid. Of course, you need to consult with an appropriate lawyer if you really want to know whether or not you need to pay...
> its better to learn the discipline first
Sure, if these kids are going to become professional programmers. But if this class is intended to just give them a taste of programming or some simple skills - they discipline isn't the most important thing to learn.
Actually accomplishing something is important. And doing it with a minimum of distraction and fuss is also probably important. And if this is the case, then python is probably a nearly perfect first language. And since python tends to push you towards a right way of doing things, you will tend to pick up some valuable discplines there anyway.
I suppose you could also argue that VB might also apply since it's all about just getting things done, but it's so ugly that I think it'll tend to wreck the students for any other languages later on.
Actually, they showed a few different things:
:-)
- assembler source code
- core dump
- cobol source code
but i remember the cobol the most
You know, the first language I learned way back when was COBOL. I didn't love it much, but that's only because nobody ever told me that you could create a Terminator with it!
Of course, even as a junior programmer I probably would have been sharp enough to send information directly to the brain on the cyborg rather than just doing a printout to the eye. But you know how it goes - machine generated code is always crap.
> The old one -- even the Democratic members -- did not.
Note that much of the support for the erosion of our personal liberties occured while this country was in a patriotic & fearful frenzy. Anyone in congress that took a moment to say, "um, could we talk about this for a minute?" was attacked by:
- president bush & his administration
- senate & house leaders
- media pundits
- etc
- about 50 million americans
- right wing talk radio & tv (fox news)
All of which was enabled by most mainstream news media. Everyone was so interested in "working together" that they allowed the administration and their colleagues to manipulate everyone to pass this crap.
So, yeah many Democrats supported these intrusive policies, bills, etc. But often only because to attempt a sane action would have been the equivilent of spitting in the wind - and killed their chances to survive the next election - and eventually reverse this crap.
Remember the first invasion of iraq? When a large number of democrats pushed back? The Republicans often remind Democrats of how almost every one of those Democrats that pushed to avoid war lost their next elections. So, this time around the Democrats played it safe. Unfotunately, as much as I wish it wasn't true, that was probably the best policy for a minority party without house, senate or administration.
Of course, they could also have been doing it because they wanted government control or somehow benefitted from supporting it. But I think these latter reasons were the lesser.
It's interesting that some people think that the military spends efficiently but will then criticize the rest of government for being inefficient.
The reality is that they're all just massive bureacracies that waste money like mad - because everything takes too many review steps, too many approval steps and effective requirements gathering doesn't work this way.
> throwing out our perfectly good 2 year old body armor that we spent billions on.
well, there ya go - if you just spent billions on 2 year old, perfectly good body armor and find yourself having to throw it out - somebody made a mistake, huh?
I've been in the military, have seen a lot of money wasted - primarily due to process problems. For example, a small box of monitor 'wipes' was ordered (this was 20+ years ago). What appeared was a 700+ lb massive crate of thousands of individually-wrapped wipes. Total cost was something like $25k. These were just used as paper towels since it would have been too difficult to send it back.
> When talking about risks and countermeasures, it pays to be specific.
So, when talking to a bunch of slashdotters, the extra detail implied by the acronym 'IED' vs 'booby-trap' (or even just 'trap') has some kind of pay-off?
Like, some slashdotter out there may die trying to detect trip wires on a non-IED bobby-trap in his back yard?
Seriously, unnecessary precision in your speach can cause more confusion than it resolves.
> Remember how this started out - a sexual harassment lawsuit. I forget the details
You unfortunately then forgot about how the entire sexual harassment suit was funded and backed by an extreme far-right interest group. Not a group that historically has been concerned about women's rights (far from it), a group that sought to use any possible excuse to harass Clinton.
And sure, having sex with interns isn't good judgement, and everyone should be concerned with him about that. But frankly, I'm more concerned about one party disabling the government while creating a circus over a primarily private sexual relationship than I am about the president lying about who he has sex with. Lying about sex has no impact on the government, distracting the government and blowing millions of dollars to attack the president over a non-issue clearly does.
> Oracle is the shiznit when it comes to high performance general database work. It will scale far beyond almost everything else,
> with DB2 a close #2. Niche players like TeraData have their place too, but only Oracle can scale across the entire enterprise.
Sure, if you're talking transactional systems (like airline reservations). But if you're talking about data warehousing, very large scale analytics - then db2 and teradata have the upper hand. Oracle's clustering is for failover, db2's clustering is for performance.
Many years ago I was a programmer in the marine corps - and our office was in the top of an aircraft hanger. The interruptions could include a sergeant that wanted you to empty waste baskets, carry some boxes around - or an A4, F4 or maybe c-130 racing up its engines directly outside of our door.
So, ear-protection was standard issue - and we simply got accustomed to working at our desks with this equipment on. And we got work done.
Years later, I've worked in all of the environments described here, and I prefer open-plan with a small number of coworkers (7 or less). A common understanding of the "rules" is required (no music, no hollaring, etc, etc) as well as some maturity from the participants. But this beats the pants off offices in my book - in which you deliberately sever communications between team members.
This is especially true if at the same time you're wondering how you're going to implement methodologies such as pair-programming: with two people per office you typically don't invite a third person in your office for daily programming sessions. So, where exactly would this occur? In some additional conference rooms? Unused offices? Well, assuming an infinite budget for real estate I suppose this is possible - but in the real world there is seldom so much vacant space.
> Um what do you mean about the lack of command-line features? SQL Server has only one interface and that interface is SQL text sent to it from a client.
Sorry, I should have been more clear. I was referring to the limited batch processing ability of windows: that is, windows/2000 doesn't have a reasonable shell built-in. So, to automate tasks you're stuck with the windows task scheduler, primitive bat scripts, etc.
On most sql server projects I ended up installing cygwin, and running batch processing via shell scripts and python. Eventually we changed some very large applications over from using DTS to using BCP and python - with a huge and immediate reduction in maintenance effort & failures.
But most sql server dbas that encountered these systems were shocked - this just wasn't "the way" to run sql server, and these dbas just wanted to use the gui - and were obstacles to really getting tasks automated. However, the seasoned oracle, db2, etc dbas immediately recognized what we were doing and had zero problems in transitioning over. And of course when we later converted from sql server to postgresql most of the platform-independent code survived just fine.
> 1. Can be true - but depends on how small the "small" projects are, and how many there are. 15 small Oracle databases are going to be more work than 15 Postgres
> databases, even if you are already running Oracle for other stuff. Also, if there's enough work for the smaller stuff that it needs separate DBAs (even if
> it's just 2 total), then getting Postgres ones would not be more expensive than Oracle ones.
I would only go with postgresql is it really was a separate team supporting them. Otherwise the oracle dbas are now going to have to also learn how to recover postgresql databases from backup, etc, etc. That just isn't worth it most of the time.
> 2. Blatantly untrue. Oracle will scale better on multiple machines, but on a single machine Postgres will outperform Oracle far more often than not.
Not for reporting. If I've got a four-way oracle server doing reporting with 100 million rows in its fact table partitioned on day with 2-3 million rows per day. Then oracle will be far faster at most trending (reporting) queries. You won't have 365+ tables to manage for fake paritioning, you'll get 4x the performance of the single threaded open source servers, and better memory management. It'll be *far* faster than the open source options.
> And I'd like to see some source showing Oracle's optimizer to be better than Postgres'.
Well, I was really thinking about mysql here - which has historically had difficulty joining more than three tables without unpredictable performance results.
> Which version are you using?
x
SQL Server 7 & 2000
> SQL Server 2005 is MUCH cheaper than $80k for a 4 way server,
No, that depends on which version you need to use: http://www.microsoft.com/sql/howtobuy/default.msp
So, if you've got an internet-accessible search engine running sql server (wasn't my decision) then you can easily blow over $80k to license a four-way server. In fact the original estimates we got were over $100k.
> People are far more dependable when they're working for money than for charity.
not when they suck - which they frequently do when working on product support teams.
yes, I'm glad that I'm working with supported products - but I also avoid calling them like the plague. It is very much a worst-case scenario.
This is a pretty trivial article which seems driven by ingres.
Anyhow, a few things that I'd consider:
1. since as the author mentions the open source databases aren't ideal for mission critical applications (yet), then many organizations will find themselves supporting multiple databases. Say, oracle for financials & crm & the corporate warehouse and postgresql for a variety of smaller projects. Makes sense in many ways - except: oracle is already free for the small databases anyway, and now you need the dbas to support multiple products. This is going to increase your labor costs - not decrease it.
2. for many large analytical databases (data warehouses, etc) the cost of using open source are actually higher than closed source. This is because db2, oracle, etc are better at using the hardware than the open source alternatives. They've got better optimizers, parallelism, far better partitioning, better better pool management, automatic query rewrite, etc. So, a $100k oracle lisense running on a $100-200k 4-way (i know, assumes discount) will out-perform postgresql (free) on a 16-way ($1m) in many ways.
3. for some applications mysql could be more expensive than oracle. Ok, not just because you need to do far more testing with mysql to make sure that none of the wacky silent errors are affecting your code. But also because of the odd licensing - that requires its own faq and tips to just license the product if you can't figure it out. Then there's enterprise db - not very familiar with this one, but I doubt that it is free. Meanwhile, at the low-end the big-three database vendors all support free products. So, whether or not you pay more may very well depend on how you use the software.
Of course, if you're at a company like mine, and get to bypass purchasing and just review the license & install - you probably are saving a vast amount of money after all.
I actually found sql server to be quite expensive - from licensing (which was running > $80k for a 4-way on enterprise edition) to labor.
The lack of command-line features meant that many operational activities that could be automated required a dba to manually do the job via the gui. And lets not even talk about how you had to completely recreate DTS packages when promoting them from dev to test to prod...
So, there are labor savings that you can get on sql server vs oracle, db2, postgresql, etc - but the lack of a command line interface wasn't a driver in my experience.
> When a computer doing the counting, you'd expect 100% accuracy.
no, there are legitimate reasons for inaccuracy, such as:
- user selects wrong candidate (perhaps they don't line up right in the display, etc)
- user selects a candidate, then backs up and selects a different one, but does this process incorrectly
- user's vote isn't completed and so entire ballot is discarded
- date is incorrect on voting machine
- etc
These aren't hardware or software errors, they're mostly human errors or usability defects. But historically I think this is where the problems are most likely to occur and have occured with paper ballots as well.
Of course, there could be something nefarious going on. I wouldn't instantly rule that out. Just wouldn't assume that just because someone thinks that there is zero chance of error here.
>> Many precincts are too small for generators to be practical, and UPS units also have a failure rate. What if, even
... fighting for the democracy ?
>> though it was tested the week before, the generator fails on the day of the election? There is also the cost associated.
>> Who is gonna pay for it all.
> Wouldn't that be like
> Sorry, I'm not an american, but I though you people didn't mind spending money while fighting for democracy. But maybe
> I misunderstood, and all that money is for fighting for something else.
Sorry, but you're being naive. There's no way to ensure that a meteor or tornado doesn't strike, that a fire doesn't burn the building down, that the building doesn't plunge into a sinkhole, or that wild dogs don't kill everybody in the building and eat the computer.
Not likely scenarios. Neither is it likely that a well-designed computer system will fail. But they do occasionally. Hard drives fail, power supplies fail, memory fails, etc, etc, etc. Even if you go with redundant components the motherboard may crack, etc. If your objective is to create a voting machine that never fails you'll just spend billions of dollars and still only get 99.99999% availability.
So, be practical. Ensure that the system has good uptime (99.99%?), is extraordinarily difficult to hack, leaves a good audit trail, and that voting and auditing are an atomic action. That should be fine. And should be more valid than paper ballots.
LiquidCooled wrote:
> I personally think mysql handles slash really nicely and coupled with the open source foundations it stands on allows
> a lot more people to get systems up and running, moving to a proprietary database (even a free one) would exclude a lot of
> people from running their own site with slashcode.
Yep - open source products are definitely more agile than closed source - especially if you have to work through a horrible centralized procurement process like I have to. However, I wouldn't assume that mysql helps slashdot out today. Aside from today's snafu:
- why has the addition of new functionality taken so long?
- why is it so painful to see all threads by subject over the last x days, filtering by y?
- why can't we easily see better metrics on threads?
- why can't we easily see better metrics on posters?
- etc
Well, one reason is that the database would have to return aggregate results for many queries - and mysql is a bad match for mild analytical queries - in which you need to scan 10k rows and return fifty.
So yeah, mysql was about the best option five years ago, but has had commercial options almost as inexpensive for the past two years and open source options with better data quality and just as good performance for the past three years or so.
> Maybe there are enough database developers out there today who have been effected by this who could step up to the plate
> and fix mysql itself so it does have the required features. There are certainly enough geeks in here.
Well, mysql has been getting much better over the years: just a few years ago:
- it had no transactions (and their vps said this was a good thing)
- no views
- no triggers
- no stored procedures
- no subselects
- a trivial optimizer (5-way joins were almost unrunnable)
- silent errors on data truncations (truncates numbers)
- silent errors on data types (allows feb 58th, 2004)
- silent errors on transaction-safe storage creation
it's far better today, but still has inexcusable problems - in which a simple column addition requires an entire table to be rebuilt, adding an index to a 16m row table takes 3+ hours (!?), you can turn on error reporting for type failures, etc - but any client can override the server setting, etc.
So, they've come a long way, but are still only halfway there. The question for developers that want to jump in and help fix things - is where is their time better spent? Helping a for-profit company fix its product, or just using a completely unencumbered and free product like postgresql in which these kind of errors have never existed.
LizardKing wrote:
> I guess this is another thing to add to the MySQL gotchas page [sql-info.de]. Of course, in a decent database engine,
> like PostgreSQL, if you alter a column data type then the indexes are updated to reflect this.
Nah, this is a feature - each storage engine for mysql can do it differently. Sometimes they might behave like any other database, at other times you might have extra steps. And you can even get this benefit within a single database across multiple tables.
FYI, regarding partitioning as a design consideration:
- Yes, it is well-worth looking at, but note that partitioning isn't as fast as btree indexes for retrieval - *if* you were only planning to grab a few rows. It's more of a reporting thing that helps save you from table scans when you need to grab thousands of rows.
- Also, I wouldn't necessarily trust the merge table or partitioning features out of mysql until others have worked out the kinks first: there's a lot of potential complexity there with how indexes work (global or local), how the optimizer works, etc.
- And you probably will have sql changes and maybe column changes - in order to incorporate something like a story-id, thread-id or time-id to partition upon and to use as a query predicate.
Regarding your product selection - if you're going to look at changes for partitioning anyway:
- it wouldn't hurt looking at a database with fewer sharp edges to injure yourself on either. I mean, really - you've got degraded functionality right now because of design defects in the tool
- which allowed the error to occur and take so long to reindex and process defects in that your process doesn't require the additional testing that is essential when using dangerous tools like mysql.
- side note: if you were using a database engine capable of reindexing quickly in parallel you would have had that index built in probably a fraction of the time. Sometimes it's worth spending a few dollars to get infrastructure that you can really depend on. Doesn't have to be much cash, I think the free version of db2 (limited to 2 cpus & 4 gbytes of memory) would probably be a substantial improvement.
We're using multiple methods here:
1. ETL (extract/transform/load) processing in which source data is translated to the warehouse format. This consists of several components:
a. python-based etl framework - responsible for managing our processing queues. This code takes a decentralized approach quite different than most commercial etl tools - that are large and awkward monolithic applications. This has worked very well.
b. python-based transformation - typically simple custom-written programs that read a single file and write a single transformed file. Execution of these programs is handled by the etl framework - based on existance of data in the queue. Lookups are via python hash tables. This has proved to be an extremely easy to develop and maintain solution.
c. python-based, metadata-driven validator and transformer - warehouses are easy to load quite a lot of data quickly into (hitting easily over 100k rows / second) but very slow to delete data from. Additionally, many of our source systems will change business rules without informing us. So, we have a completely separate python validator that is handed each field individually, then writes the output row. It assumes that all business rules are handled by the calling transformation program - so it only double-checks type, date format, value ranges, default values. It also is ultimately responsible for final file formatting - this is because due to so many control characters in my data I need to go with fixed-length files, and formatting so many fields this way was so error-prone.
2. testing framework - we've got a few testing challenges in the warehouse:
- server configuration consistency between multiple read-only servers
- data consistency between sources and targets and between multiple read-only servers
- data quality due to lack of enforced foreign key & check constraints
- configuration policy compliance (corporate security and departmental standards)
- performance of complex queries
Our tester generates tests for many of the above, and allows us to easily run tests for all of the above. We use it in both development for QA and then daily for continual QC (emails us alerts if there are any).
> It is just god-awful slow.
Really? hmmm, my team runs a huge data warehouse in which all data is transformed via python programs. This means 40+ million rows a day get hash table lookups, business rule validations, type changes and various transformations.
Then most of that data goes through an additional metadata-driven validation process that checks and confirms every single field. Again, metadata-driven - so a oouple hundred lines of various hash table lookups to determine exactly what validation applied to each field.
Ok, now let's talk performance:
1. io performance (most critical factor here)
a. same as c, in other words, extremely fast
2. cpu performance
a. much slower than c, but workable
b. psycho helps a ton
c. the ability to rewrite key functions and classes is cool
Let's also talk about productivity
1. great for ease in training part-time or junior programmers
2. great for immediately understanding old code
3. great for writing robust test-driven code
4. good for very quickly assembling a solution
The above combo of ease of writing, ease of maintaining puts it far ahead of perl, tcl and java. Ruby is really the one other one in this niche I can think of off the top of my head. And even there Python appears stronger on the maintenance side.
So, yeah - I'm sure there are many applications that have more significant performance requirements than ours does. But for this large application it is completely fine.
> Respected by those who agree with them, yes. Vilified by those who don't. The "Journalists" of Foxed News are probably even
> more respected by more people. Why not then force taxpayers to subsidize Fox News as well?
Even most conservatives I know listen to NPR. I'm sure there are some of the Rush Limbaugh type who can't stand it. But keep in mind as a liberal I think it often reflects a much more pro-administration and conservative viewpoint than I do. Those that vilify it might be angered by the fact that they do give airtime to different views. That's got to really burn them up.
> If the forced money is so small, then they can do without it, right?
Sure, punish them for being affordable while letting the military blow billions on pork-barel. That would be a great deal.
> The majority of which is CLASSICAL music. Which was creative and innovative...a couple of hundred years ago.
ah, you probably don't listen to public radio. Note that NPR != public radio, it is a set of services and syndicated programs oftened subscribed to by public radio stations. In my town I've got at least two public radio stations - one just plays classical music and is completely unaffiliated with NPR. The other is affiliated with NPR but plays no classical music. It does play local music, blues, jazz, worldbeat, bluegrass, celtic, zydeco and rock. Unlike a mainstream station it has a playlist as long as whatever the dj wants to play - instead of just 10-20 tunes.
Of course, your mileage may vary - but I've found pretty high quality of programming at all public radio stations that I've encountered so far.
On the other hand, if all you want is an extreme conservative point of view and top 10 country hits, then public radio probably isn't for you.
> If NPR wants to beg for money from listeners, it should have to compete in the open market. I shouldn't be forced to support them, at any amount.
Hmm, I'd think you would want all projects to work like public radio stations: they merely get seed money from the government, then have to augment that with voluntary contributions from listeners. That ends up taking around two weeks a year for my local station.
So, what you get from public radio:
- some of the most respected and professional journalism available in the US
- some of the most creative music across all genres available in the US
- high quality programming (Praerie Home Companion, This American Life, Wait Wait Don't Tell Me, etc)
- local news
And what does it cost you? probably a few pennies a year in gov taxes plus whatever you care to donate.
Now, imagine if the military or some other truly huge government activities were funded this way! Don't like our boondogle in Iraq? Great, not only can you hopefully vote the bums out, but you can also withhold revenue. Having spent years in the military I can tell you that it is much less professionally run than my local *public* radio station!