The reason all SELECTS should be SELECT DISTINCT only referred to my example. In the relational model, all elements are distinct. The tables are sets, not multisets(or bags). So mapping from relational algebra to SQL, all results should only contain distinct elements.
For instance, projecting a single column from a table in the relational model will only result in the set of distinct elements from that column. In SQL, doing the same will result in one element per input row, which may mean multiple copies of some elements.
Your object database is a relational database with objects.
There are object-oriented databases. I'm not 100% sure that object-store is one of them, but oo databases are definitely not relational databases. For one thing objects have identity external to their data. In the relational model, the ONLY way of identifying a relation(aka row) is by the data it contains. Additionally, relationships between objects are viewed as links/references to objects, not foreign keys. oo databases allow for methods inside objects. The fact that the storage usually uses a relational DB as the back end is irrelevant to the programmer. He doesn't care if flat files are used either as long as things are fast.
Just because something uses a relational DBMS as a back-end doesn't mean it doesn't provide a model that is not relational. That would be like saying that a relational DBMS isn't any different than a flat file because flat files(or block devices) are that back end used to store the data.
The associative part is simply a program that uses a relational database to hold its data.
I think I may have used the wrong term there. I was referring to things like the berkeleyDB, which provides an interface to associate keys and values, but not much more. Basically a persistent lookup table
That may be, but I haven found few operations easier to express in relational algebra than in SQL.
for example:
I want the name field from a if it's id is in b.
In relational algebra, PROJECT[A.name](THETA_JOIN[A.id=B.id](A, B)) or p[A.name](A |X|[A.id=B.id] B) Sorry, ascii sucks.
In SQL SELECT DISTINCT A.name FROM A,B WHERE A.id=B.id;
I find the SQL version to be more readable, etc. The same functionality is provided by both and is easily transferable.
cartesian production becomes SELECT * FROM A,B
natural join becomes (A NATURAL JOIN B)
theta join becomes SELECT DISTINCT * FROM A,B WHERE
selection becomes SELECT * FROM A,B WHERE...
projection becomes SELECT columnA, columnB FROM A,B
With nested queries, everything is easily translatable from relational algebra to SQL (Technically, all SELECTS should be SELECT DISTINCT, but whatever). Otherwise, temporary tables can be used.
The real reason relational algebra seems easier to deal with is because you can use symbols to represent things and there are no data types. It is an abstract language that cannot be implemented because it is under-defined.
If you try expressing a hairy SQL query with relational algebra syntax, you will end up with a hairy relational expression.
In corporate deveopment, the "bandwagon mentality" is not always a bad thing. If you are using what everyone else is using and you hire someone new, they can easily start right away with development.
If you think joins suck (I don't personally agree, but that's all just opinion), you should be using some other database model. After "data is stored in a table", joins are THE feature of relational databases.
Maybe try an object-oriented database (I hear ObjectStore is good), or an associative database (BerkeleyDB), or an XML database (Sleepycat has one).
Relational databases aren't the only game in town, they are just the most popular and therefore:
have had the most development time thrown at the DBMSs
While we are attacking statistical rigor, you have assumed that the distribution will be heavy in the middle like a bell curve (probably reasonable, but not always). If the distribution is uniform, the mean will indeed shift to a 3.0 and if it is heavy on the ends, it would shift above a 3.0.
The assumption is probably valid for most institutions, but might not be depending on the grading policies in place. For instance, I would imagine that the distributions for some graduate schools that require a B or better for a class to count would be skewed toward the high end because a lot of C's might be bumped up to B's, but a professor wouldn't bother bumping a high D up to a C because it doesn't make any difference.
In the case of IQ's, it is probably safer to assume a gaussian distribution because it is an open ended score. If you close the ends and give IQ as LOW, AVG, and HIGH, where LOW means 101, there would probably be a nearly uniform distribution among the three scores (I'm not an IQ expert, the numbers may need to be different, but you see the point).
I thought it was evident from the post, but I agree that gov't should keep their dirty fingers out of our business. The basic skills should be required by employers and (through demands for higher quality) those employers' customers.
I also agree that exams and pieces of expensive paper do not necessarily mean someone has any skills at all (other than a skill at acquiring expensive paper). However, many employers seem to think that programmers are unskilled labor and they hire accordingly. This results in products comparable to a cabinet I built instead of a cabinet built by a skilled cabinet maker. And end users have come to expect poor quality, so they do not demand any different.
Maybe those skills aren't absolutely required, but there a basic skills that people should have to get any job in a particular field whether all of those skills are strictly required or not.
All civil engineers should be able to calculate the weight a concrete slab can support even if they are hired only to do surveying and planning a dirt road.
An electrical engineer should be able to look at a schematic and tell the potential difference between any two points even if he is being hired to do logic design at the gate level.
People have grown so used to crappy software they expect it and they know they can get it from unskilled programmers. Programmers are a dime a dozen because the market puts low emphasis on quality.
I'm not saying a degree or even a certification is a prerequisite for being a good programmer, just like it isn't a prerequisite for being a good electrical engineer, but in all fields there is a base level of skill that should be required.
Many people use 'radio waves' to denote any communication using wireless em not in the visible range, even if it isn't technically the right frequency to be radio waves.
Actually, IBM didn't hold on to the PC market because they didn't use proprietary crap. The reason other makers could make a PC-compatible system was because IBM used off-the-shelf components. The only proprietary part on the PC was the BIOS and it was easily reverse engineered.
This isn't at all what they are trying to do. It is the exact opposite. Sun is trying to create a technology that raises the bandwidth the processor has with the outside world to the point that things like cache can be REMOVED from the chip, not added to it.
The immediate effects would be smaller on-chip caches and faster communications between processors on multi-processor machines.
This is not a problem with the SQL, but with the coding. It makes things very easy to read and more easily maintainable if you isolate each SQL query/command or small sets of queries/commands in their own functions/class.
Not only does this make the main logic more readable, it allows you to modify the SQL implementation without touching a lot of code.
For instance if you split one table with a column with a few large pieces of data that are repeated many times into two tables, one with a reference to a key for the large data pieces. Or if you add an index that makes a different SELECT statement faster than the current.
Not really, this is more like the alimony giver having the ability to pay in cash, check, money order, or some other currency.
This is more like the courts saying that the "giver" must give 3000 pieces of paper currency a month, assuming that the pieces of currency will be USD, but they then give 3000 1 Turkey Lira notes instead. Which computes to 0.2 cents US.
Then there is the other problem that many people seem to be ignoring that ethernet is by design limited to a pretty short distance (I'm too laxy to pull out the networking book). And the fact that because you might have a 10Gbps connection to one other computer, doesn't mean you are gonna have a 10 Gbps connection to anyone else. I know I have 100Mbps within my apartment, but then there is that darn internet connection that tops out at ~2Mbps.
In your world, 'hdtv' might be 1280x1024, but in the world at least a few of us live in, HDTV (notice the caps) isn't, it's 960x540(assuming square pixels) and I imagine less than 24 bit color (All this analog stuff!). I don't know, but I would also guess that it's the same frame rate as regular old TV (~30 fps).
906 x 540 x 16 x 30 = 235 Mb add sound and round up give 250 Mb and forty channels, with sound. Still not stupendous, but even with some fast, lossless compression that would probably double to 80.
I second that. If you really "know linux inside and out", you would know that you can install things in your home directory.
Being an admin for a developer lab, I also know things sometimes take longer than they should because IT people are overworked. "It shouldn't take more than a few minutes for you to reset my password and restore my home directory because I erased everything and then forgot the password." Well, no it shouldn't if there weren't already 15 other users who did exactly the same thing this morning as well as this annoying scientist who wants me to install some weird software package I've never heard of.
One way you might want to hurry up the IT guys for installing software is give them an rpm/deb and say here, this is ??? I need it installed on ???, I downloaded it from ???. Or if they are a little more concerned with security, here is a url for the rpm I want installed, or here is a spec file for the rpm I want installed, get the source from... While I know pachages aren't required to install most software, it does make administration soooo much easier, especially if you need the same thing on multiple machinec and for recovery if the software gets lost somewhere.
There is a big difference between ht and multi-core. With multi-core, you actually have duplicates of all the processing components on chip and the performance is similar to that of two seperate processors (on-chip cache sharing and faster interconnects aside).
However, your post leads into a point I considered making already. When/If licenses do change to take into account for dual core chips, how will they address hyperthreaded processors where only bits and pieces of the core are duplicated? Should these count as 1, or 1.3, or 2 processors?
They really need to get to the point where people instantly think "Linux company" instead of "failed network company" when they hear the Novell name.
Did you read the article?
No, I'm not new here, but the article discusses exactly this point.
They don't want to be known as a Linux company or an open source company because they don't plan on abandoning their NetWare customers or at least they don't want their NetWare customers to abandon Novell.
This is probably a good idea. I know if I were to run a software company that happened to be based around Linux, I wouldn't want to be known as an OS company because of the connotations of zealotry that come along with the moniker. I would want to be known for using the right tool for the job, be it open source or proprietary or written in house under whatever licence you like.
I happen to be a software writer (call me an engineer or a programmer, whatever you like).
In my field, I have two ideas that are somewhat related.
1. Create a certification program for various software disciplines. It should be by engineers/programmers for engineers/programmers. It should be free (as in speech) and as close to free (as in beer) as possible. Possibly developed using a model where certified practitioners give feedback and continue to contribute to the test as part of their continued certification. I know there are many problems with this, but I would love to see a better system.
2. Form a union. Please don't flame me, I'm a strong conservative and against most incarnations of unions, but they are created for situations just like our own where the workers in an industry are not being treated appropriately by the employers.
I would completely disagree with wage negotiation on the part of a union, however. I would see their primary role as promoting certification to both engineers and employers.
I'm by no means happy with this suggestion, but I thought I'd throw it out there. The central point I would like to make is that (in the software and IT industry) the wages are low and we get little respect because there are a whole lot of very unqualified, unprofessional, and unproductive people in the field and employers have gotten to the point that low productivity and poor quality are what's expected, lower wages which further lowers moral, productivity and quality.
Slashdot Mirror
The reason all SELECTS should be SELECT DISTINCT only referred to my example. In the relational model, all elements are distinct. The tables are sets, not multisets(or bags). So mapping from relational algebra to SQL, all results should only contain distinct elements.
For instance, projecting a single column from a table in the relational model will only result in the set of distinct elements from that column. In SQL, doing the same will result in one element per input row, which may mean multiple copies of some elements.
Your object database is a relational database with objects.
There are object-oriented databases. I'm not 100% sure that object-store is one of them, but oo databases are definitely not relational databases. For one thing objects have identity external to their data. In the relational model, the ONLY way of identifying a relation(aka row) is by the data it contains. Additionally, relationships between objects are viewed as links/references to objects, not foreign keys. oo databases allow for methods inside objects. The fact that the storage usually uses a relational DB as the back end is irrelevant to the programmer. He doesn't care if flat files are used either as long as things are fast.
Just because something uses a relational DBMS as a back-end doesn't mean it doesn't provide a model that is not relational. That would be like saying that a relational DBMS isn't any different than a flat file because flat files(or block devices) are that back end used to store the data.
The associative part is simply a program that uses a relational database to hold its data.
I think I may have used the wrong term there. I was referring to things like the berkeleyDB, which provides an interface to associate keys and values, but not much more. Basically a persistent lookup table
That may be, but I haven found few operations easier to express in relational algebra than in SQL.
, B))
...
for example:
I want the name field from a if it's id is in b.
In relational algebra,
PROJECT[A.name](THETA_JOIN[A.id=B.id](A
or
p[A.name](A |X|[A.id=B.id] B)
Sorry, ascii sucks.
In SQL
SELECT DISTINCT A.name FROM A,B WHERE A.id=B.id;
I find the SQL version to be more readable, etc. The same functionality is provided by both and is easily transferable.
cartesian production becomes SELECT * FROM A,B
natural join becomes (A NATURAL JOIN B)
theta join becomes SELECT DISTINCT * FROM A,B WHERE
selection becomes SELECT * FROM A,B WHERE
projection becomes SELECT columnA, columnB FROM A,B
With nested queries, everything is easily translatable from relational algebra to SQL (Technically, all SELECTS should be SELECT DISTINCT, but whatever). Otherwise, temporary tables can be used.
The real reason relational algebra seems easier to deal with is because you can use symbols to represent things and there are no data types. It is an abstract language that cannot be implemented because it is under-defined.
If you try expressing a hairy SQL query with relational algebra syntax, you will end up with a hairy relational expression.
In corporate deveopment, the "bandwagon mentality" is not always a bad thing. If you are using what everyone else is using and you hire someone new, they can easily start right away with development.
It's all about maintainability.
Sounds like a one to many relationship to me. Or possible a many to many, it just changes what the keys are.
Each dynamic column becomes a row in a second table.
table a
id name syllables
1 foobar foo, bar
translates to:
table a
id name
0 foobar
table syllables
a_id syllable
0 foo
0 bar
If you want ordering, you can add another column to syllables to represent order.
Maybe try an object-oriented database (I hear ObjectStore is good), or an associative database (BerkeleyDB), or an XML database (Sleepycat has one).
Relational databases aren't the only game in town, they are just the most popular and therefore:
Shhh!!
It doesn't work if you tell them the plan.
Sorry, no better here, don't bother with that trans-Atlantic migration you had been planning.
We use three 1TB usb/firewire drives at work.
While we are attacking statistical rigor, you have assumed that the distribution will be heavy in the middle like a bell curve (probably reasonable, but not always). If the distribution is uniform, the mean will indeed shift to a 3.0 and if it is heavy on the ends, it would shift above a 3.0 .
The assumption is probably valid for most institutions, but might not be depending on the grading policies in place. For instance, I would imagine that the distributions for some graduate schools that require a B or better for a class to count would be skewed toward the high end because a lot of C's might be bumped up to B's, but a professor wouldn't bother bumping a high D up to a C because it doesn't make any difference.
In the case of IQ's, it is probably safer to assume a gaussian distribution because it is an open ended score. If you close the ends and give IQ as LOW, AVG, and HIGH, where LOW means 101, there would probably be a nearly uniform distribution among the three scores (I'm not an IQ expert, the numbers may need to be different, but you see the point).
Excellent comment.
It seems that the point needs to be made for some of us who think we know more than we do.
Dynamite == TNT == Nitroglycerin in Fuller's earth. Invented by Nobel (Think the prize). Highly explosive. Sweats nitroglycerin.
I thought it was evident from the post, but I agree that gov't should keep their dirty fingers out of our business. The basic skills should be required by employers and (through demands for higher quality) those employers' customers.
I also agree that exams and pieces of expensive paper do not necessarily mean someone has any skills at all (other than a skill at acquiring expensive paper). However, many employers seem to think that programmers are unskilled labor and they hire accordingly. This results in products comparable to a cabinet I built instead of a cabinet built by a skilled cabinet maker. And end users have come to expect poor quality, so they do not demand any different.
Maybe those skills aren't absolutely required, but there a basic skills that people should have to get any job in a particular field whether all of those skills are strictly required or not.
All civil engineers should be able to calculate the weight a concrete slab can support even if they are hired only to do surveying and planning a dirt road.
An electrical engineer should be able to look at a schematic and tell the potential difference between any two points even if he is being hired to do logic design at the gate level.
People have grown so used to crappy software they expect it and they know they can get it from unskilled programmers. Programmers are a dime a dozen because the market puts low emphasis on quality.
I'm not saying a degree or even a certification is a prerequisite for being a good programmer, just like it isn't a prerequisite for being a good electrical engineer, but in all fields there is a base level of skill that should be required.
Even if that happens, greenhouse gases will be 70% of 130% (or 130% of 70%), which is still a 9% reduction.
Many people use 'radio waves' to denote any communication using wireless em not in the visible range, even if it isn't technically the right frequency to be radio waves.
Actually, IBM didn't hold on to the PC market because they didn't use proprietary crap. The reason other makers could make a PC-compatible system was because IBM used off-the-shelf components. The only proprietary part on the PC was the BIOS and it was easily reverse engineered.
This isn't at all what they are trying to do. It is the exact opposite. Sun is trying to create a technology that raises the bandwidth the processor has with the outside world to the point that things like cache can be REMOVED from the chip, not added to it.
The immediate effects would be smaller on-chip caches and faster communications between processors on multi-processor machines.
This is not a problem with the SQL, but with the coding. It makes things very easy to read and more easily maintainable if you isolate each SQL query/command or small sets of queries/commands in their own functions/class.
Not only does this make the main logic more readable, it allows you to modify the SQL implementation without touching a lot of code.
For instance if you split one table with a column with a few large pieces of data that are repeated many times into two tables, one with a reference to a key for the large data pieces. Or if you add an index that makes a different SELECT statement faster than the current.
Not really, this is more like the alimony giver having the ability to pay in cash, check, money order, or some other currency.
This is more like the courts saying that the "giver" must give 3000 pieces of paper currency a month, assuming that the pieces of currency will be USD, but they then give 3000 1 Turkey Lira notes instead. Which computes to 0.2 cents US.
Well, I get 40, but whatever.
Then there is the other problem that many people seem to be ignoring that ethernet is by design limited to a pretty short distance (I'm too laxy to pull out the networking book). And the fact that because you might have a 10Gbps connection to one other computer, doesn't mean you are gonna have a 10 Gbps connection to anyone else. I know I have 100Mbps within my apartment, but then there is that darn internet connection that tops out at ~2Mbps.
In your world, 'hdtv' might be 1280x1024, but in the world at least a few of us live in, HDTV (notice the caps) isn't, it's 960x540(assuming square pixels) and I imagine less than 24 bit color (All this analog stuff!). I don't know, but I would also guess that it's the same frame rate as regular old TV (~30 fps).
906 x 540 x 16 x 30 = 235 Mb
add sound and round up give 250 Mb and forty channels, with sound. Still not stupendous, but even with some fast, lossless compression that would probably double to 80.
I second that. If you really "know linux inside and out", you would know that you can install things in your home directory.
... While I know pachages aren't required to install most software, it does make administration soooo much easier, especially if you need the same thing on multiple machinec and for recovery if the software gets lost somewhere.
Being an admin for a developer lab, I also know things sometimes take longer than they should because IT people are overworked. "It shouldn't take more than a few minutes for you to reset my password and restore my home directory because I erased everything and then forgot the password." Well, no it shouldn't if there weren't already 15 other users who did exactly the same thing this morning as well as this annoying scientist who wants me to install some weird software package I've never heard of.
One way you might want to hurry up the IT guys for installing software is give them an rpm/deb and say here, this is ??? I need it installed on ???, I downloaded it from ???. Or if they are a little more concerned with security, here is a url for the rpm I want installed, or here is a spec file for the rpm I want installed, get the source from
Sounds like it's not the server, but the browser. Some javascript in a page is upsetting it.
More info would be helpful, unless of course this is a joke I'm not getting.
There is a big difference between ht and multi-core. With multi-core, you actually have duplicates of all the processing components on chip and the performance is similar to that of two seperate processors (on-chip cache sharing and faster interconnects aside).
However, your post leads into a point I considered making already. When/If licenses do change to take into account for dual core chips, how will they address hyperthreaded processors where only bits and pieces of the core are duplicated? Should these count as 1, or 1.3, or 2 processors?
They really need to get to the point where people instantly think "Linux company" instead of "failed network company" when they hear the Novell name.
Did you read the article?
No, I'm not new here, but the article discusses exactly this point.
They don't want to be known as a Linux company or an open source company because they don't plan on abandoning their NetWare customers or at least they don't want their NetWare customers to abandon Novell.
This is probably a good idea. I know if I were to run a software company that happened to be based around Linux, I wouldn't want to be known as an OS company because of the connotations of zealotry that come along with the moniker. I would want to be known for using the right tool for the job, be it open source or proprietary or written in house under whatever licence you like.
I agree whole-heartedly.
I happen to be a software writer (call me an engineer or a programmer, whatever you like).
In my field, I have two ideas that are somewhat related.
1. Create a certification program for various software disciplines. It should be by engineers/programmers for engineers/programmers. It should be free (as in speech) and as close to free (as in beer) as possible. Possibly developed using a model where certified practitioners give feedback and continue to contribute to the test as part of their continued certification. I know there are many problems with this, but I would love to see a better system.
2. Form a union. Please don't flame me, I'm a strong conservative and against most incarnations of unions, but they are created for situations just like our own where the workers in an industry are not being treated appropriately by the employers.
I would completely disagree with wage negotiation on the part of a union, however. I would see their primary role as promoting certification to both engineers and employers.
I'm by no means happy with this suggestion, but I thought I'd throw it out there. The central point I would like to make is that (in the software and IT industry) the wages are low and we get little respect because there are a whole lot of very unqualified, unprofessional, and unproductive people in the field and employers have gotten to the point that low productivity and poor quality are what's expected, lower wages which further lowers moral, productivity and quality.