yeah, I don't like the dbdebunk stuff either. they spend too much time talking about how ignorant everyone is and too little time talking about real practical solutions.
> [schnip] and you don't have source to some of them.
if I don't have the source to applications that are accessing my database, then I'm in pretty big trouble. not having the source means I don't trust the app, and at that point it's like having a wide open door to your db. Sure you could play catch up with some triggers and such, but really, you're already in a bit of a mess aren't you?
Let's assume a reasonable environment though, we're you're building an app, and have control over the source. So - if twenty apps update 5 tables, then which sounds better - leave the database wide open and assume they can be updating it any which way, biting your nails worrying about a data integrity problem (like - maybe application a updates column a in one way and application b does it in another) or put the update code in one place, such that all applications go through the same api.
Preferably this api is a middle ware api, but it could just be an object to, middle ware is easier to maintain, but perhaps has more up front infrastructure.
As far as performance goes, that's just total bullshit. If anything, moving the core integrity checking off to another host, and freeing you're database server to do record management only will speed things up. Further, if reads and writes are funneled through a correct api you only need to change one set of code once when you want to change database schemas, which you _will_ want to do. Business requirements change, tables need to be added.
btw - in you're description of what 'professionals' do, how do the programers ever know of the logical data model if they are hitting the physical data directly?
I was waiting for someone to point out that the given example of a subselect is really just a simple join. That's really just SQL 101. Now, there are cases for subselects, but they do get hairy.
btw - middle tier all the way - what do you do in sql when you want to say, check db value a, based on return of C function b? run home to mom that's what. middle tier promotes code re-use, but it also promotes a _logical_ data schema over the _physical_ data schema which is exactly what putting logic in things like triggers and such in does not do.
any post that starts with, 'this will likely get moderated down because of the slashdot bias' should get modded down.
it's just throwing mud, why not discuss the article? for the record, while slashdot might have an anti-ms bias (_might_) they probably give ms more press and publicity than any non-ms site I know of. consider the frequency of days where there are 3 posts dealing with microsoft (like today).
seriously - check out www.mini-itx.com for some greate case mods. they use the low power, how heat via mini-itx motherboards and look perfect for interesting case mod projects. not high power work stations, but just right for pretty media boxes or advanced routeres, even a nice quiet workstation.
I know this article is a joke but there is a very real solution to the noise and form factor problems. It's via's mini-itx motherboards.
Check out mini-itx.com the boards are low heat, low noise and fairly powerful (500-700 mhz). They are perfectly suitable for a media box, router, or low end linux/windows workstation. All parts included they whole things comes in at under $500 usd. That's pretty reasonable, and the site has a greate many interesting links about projects (like the one I think is cool, the computer in a cigar-box, the 'Hummidor64').
this whole - 'the corperation has many different parts and ideas...' thing is just warmed over defeatist bullshit.
here is a question - do corperations have leaders? do those leaders have real power to direct the company? anyone who has been inside a big corperation knows that this is almost always a yes. while parts may disagree, they still follow the will of those who direct them. so when a corperation takes action, don't be lulled into thinking that it represents some wing minority in the company, because it doesn't. it represents the will of the one or sometimes very few people in charge, and those people should be held accountable.
further, anyone who supports that company cannot be exused of their support under the guise of confusion or disagreement. you either support it or you don't.
do these render farms use any graphics chips or are they done entirely in cpu? the reason I ask is that linux does not have any high quality open source opengl that supports the latest graphics boards. nvidia is probably the best for opengl support but not opensource.
of course! that's why it fits in so well with the unix/internet model, because it is really just a text file, and the entire tool chain built around operating on text files can be utilized.
there isn't anything really magical about xml, it is just a way of unifying and formaziling your text data.
This whole discussion is entirely wrong in its direction. While the rest of the world is moving towards managing data in a user space, world readable, flexible format that is xml, microsoft is yet again going backwards into proprietary extensions and api's that aren't transferable.
Sure there might be some speed advantatages in certain places, but that will in no way make up for the fact that you're data will be burried deep inside the os, as opposed to freely available as it is in xml.
The progress of XPath,XQuery, Schemas and xml libraries in general makes me confident that in two years using a xml as a primary data store as well as programming interface will be a breaze. Think about it, what is really missing from xml that a relational database has right now? Basically some indexing scheme and a good api to handle locking and concurrency, other than not really a whole lot. Throw in a little client server and you're done. Now once you've gone that far, what does an object data base have that an xml database doesn't? Not a whole lot, throw in some XPointer stuff, and you've got references nailed.
Pretty much anything that can be locked away in an os can be done better, and more flexibly in user space. That is why unix is better than vms, multics, or windows or whatever mainframe os, not because it has more features or higher speed, but becuase of it's light and flexible api. Files are stripped bare of anything more than the bare minimum. That keeps things flexible and easy, everything else is moved into the library.
I for one will never, ever again trust a fundamental part of infrastructure to anything closed source. This goes for stability, security, bug fixes and product lifetime.
Simply put, open source has become for serious developers a prerequisite. It means that while, yes there will be flaws, that you cannot be held hostage by these flaws be they security or functionality by the vendor. No amount of arguing statistics can change this. Developers have seen the light and will not go back, sorry.
So articles like these are really just moot. I really don't care if AIX had fewer security flaws than Linux. It only takes one. It only takes one bug that interrupts your business, and one IBM policy change, saying, 'sorry wait for the next version', or 'sorry, we don't support AIX anymore' for you to be over the barrel. This is serious business and there is a lot of money and time involved. If a vendor drops your product, or even slows down it's bug fixes you _will_ be in trouble.
> Writing your system to be platform-independant when there is no current or forecast need is also not a good idea.
that's really the whole idea behind the web. most companies just got so sick of microsoft desktop development headaches that they switched the the lions share of desktop apps to web apps.
also there is java (and inferno, but not so many people use that), tk and a few other tools that make cross platform client programming easier. it is worth the investment because platform changes happen more often than you think, even if you stick with ms.
register has been following this pretty closely.
they have a good editorial on what it would cost to ms to implement this as well (like dropping.net until the security implications are thought through)
here is the link -
http://www.theregister.co.uk/content/4/23791.htm l
Slashdot Mirror
regarded as heroes eh? did you get a raise? ... thought so.
would tend to agree - doesn't microsoft alread have a source program for universities and large corperations/governments?
yeah, I don't like the dbdebunk stuff either. they spend too much time talking about how ignorant everyone is and too little time talking about real practical solutions.
> [schnip] and you don't have source to some of them.
if I don't have the source to applications that are accessing my database, then I'm in pretty big trouble. not having the source means I don't trust the app, and at that point it's like having a wide open door to your db. Sure you could play catch up with some triggers and such, but really, you're already in a bit of a mess aren't you?
Let's assume a reasonable environment though, we're you're building an app, and have control over the source. So - if twenty apps update 5 tables, then which sounds better - leave the database wide open and assume they can be updating it any which way, biting your nails worrying about a data integrity problem (like - maybe application a updates column a in one way and application b does it in another) or put the update code in one place, such that all applications go through the same api.
Preferably this api is a middle ware api, but it could just be an object to, middle ware is easier to maintain, but perhaps has more up front infrastructure.
As far as performance goes, that's just total bullshit. If anything, moving the core integrity checking off to another host, and freeing you're database server to do record management only will speed things up. Further, if reads and writes are funneled through a correct api you only need to change one set of code once when you want to change database schemas, which you _will_ want to do. Business requirements change, tables need to be added.
btw - in you're description of what 'professionals' do, how do the programers ever know of the logical data model if they are hitting the physical data directly?
thank you !
I was waiting for someone to point out that the given example of a subselect is really just a simple join. That's really just SQL 101. Now, there are cases for subselects, but they do get hairy.
btw - middle tier all the way - what do you do in sql when you want to say, check db value a, based on return of C function b? run home to mom that's what.
middle tier promotes code re-use, but it also promotes a _logical_ data schema over the _physical_ data schema which is exactly what putting logic in things like triggers and such in does not do.
any post that starts with, 'this will likely get moderated down because of the slashdot bias' should get modded down.
it's just throwing mud, why not discuss the article? for the record, while slashdot might have an anti-ms bias (_might_) they probably give ms more press and publicity than any non-ms site I know of. consider the frequency of days where there are 3 posts dealing with microsoft (like today).
seriously - check out www.mini-itx.com for some greate case mods. they use the low power, how heat via mini-itx motherboards and look perfect for interesting case mod projects. not high power work stations, but just right for pretty media boxes or advanced routeres, even a nice quiet workstation.
I know this article is a joke but there is a very real solution to the noise and form factor problems. It's via's mini-itx motherboards.
Check out mini-itx.com the boards are low heat, low noise and fairly powerful (500-700 mhz). They are perfectly suitable for a media box, router, or low end linux/windows workstation. All parts included they whole things comes in at under $500 usd. That's pretty reasonable, and the site has a greate many interesting links about projects (like the one I think is cool, the computer in a cigar-box, the 'Hummidor64').
uh - no thannks, I'll stick with C for speed Java for enterprise and perl for a quick hacks.
Thanks.
done - take a look at the movie 'Manufacturing Consent' in the end is a great debate between Chompsies and William F. Buckley. Great stuff.
this whole - 'the corperation has many different parts and ideas...' thing is just warmed over defeatist bullshit.
here is a question - do corperations have leaders? do those leaders have real power to direct the company? anyone who has been inside a big corperation knows that this is almost always a yes. while parts may disagree, they still follow the will of those who direct them. so when a corperation takes action, don't be lulled into thinking that it represents some wing minority in the company, because it doesn't. it represents the will of the one or sometimes very few people in charge, and those people should be held accountable.
further, anyone who supports that company cannot be exused of their support under the guise of confusion or disagreement. you either support it or you don't.
you must have been a different page then me, because it sure looks like the same old fud to me.
do these render farms use any graphics chips or are they done entirely in cpu?
the reason I ask is that linux does not have any high quality open source opengl that supports the latest graphics boards. nvidia is probably the best for opengl support but not opensource.
bingo
tomcat->java->xml->xslt->html
> Oh by the way if you don't think Perl/CGI can handle heavy duty web sites, look at the slashcode sometime
you're kidding right?
_never_ look at slashcode
> It is just a frickin' text-based file format!!
of course! that's why it fits in so well with the unix/internet model, because it is really just a text file, and the entire tool chain built around operating on text files can be utilized.
there isn't anything really magical about xml, it is just a way of unifying and formaziling your text data.
This whole discussion is entirely wrong in its direction. While the rest of the world is moving towards managing data in a user space, world readable, flexible format that is xml, microsoft is yet again going backwards into proprietary extensions and api's that aren't transferable.
,XQuery, Schemas and xml libraries in general makes me confident that in two years using a xml as a primary data store as well as programming interface will be a breaze. Think about it, what is really missing from xml that a relational database has right now? Basically some indexing scheme and a good api to handle locking and concurrency, other than not really a whole lot. Throw in a little client server and you're done. Now once you've gone that far, what does an object data base have that an xml database doesn't? Not a whole lot, throw in some XPointer stuff, and you've got references nailed.
Sure there might be some speed advantatages in certain places, but that will in no way make up for the fact that you're data will be burried deep inside the os, as opposed to freely available as it is in xml.
The progress of XPath
Pretty much anything that can be locked away in an os can be done better, and more flexibly in user space. That is why unix is better than vms, multics, or windows or whatever mainframe os, not because it has more features or higher speed, but becuase of it's light and flexible api. Files are stripped bare of anything more than the bare minimum. That keeps things flexible and easy, everything else is moved into the library.
I for one will never, ever again trust a fundamental part of infrastructure to anything closed source. This goes for stability, security, bug fixes and product lifetime.
Simply put, open source has become for serious developers a prerequisite. It means that while, yes there will be flaws, that you cannot be held hostage by these flaws be they security or functionality by the vendor. No amount of arguing statistics can change this. Developers have seen the light and will not go back, sorry.
So articles like these are really just moot. I really don't care if AIX had fewer security flaws than Linux. It only takes one. It only takes one bug that interrupts your business, and one IBM policy change, saying, 'sorry wait for the next version', or 'sorry, we don't support AIX anymore' for you to be over the barrel. This is serious business and there is a lot of money and time involved. If a vendor drops your product, or even slows down it's bug fixes you _will_ be in trouble.
isn't this exactly what unix advocates have been telling mainframers to do all along?
plan9 is cool, but inferno is very cool - everything in plan9 but up one level into a safe vm that can run on any os, or without one!
I don't think you understand - microsoft is probably quite happy about this. it is at the very least validation of their software.
> Writing your system to be platform-independant when there is no current or forecast need is also not a good idea.
that's really the whole idea behind the web. most companies just got so sick of microsoft desktop development headaches that they switched the the lions share of desktop apps to web apps.
also there is java (and inferno, but not so many people use that), tk and a few other tools that make cross platform client programming easier. it is worth the investment because platform changes happen more often than you think, even if you stick with ms.
register has been following this pretty closely. .net until the security implications are thought through)
m l
they have a good editorial on what it would cost to ms to implement this as well (like dropping
here is the link -
http://www.theregister.co.uk/content/4/23791.ht
he he, looks like michael has egg on his face again
> She has been unable for many years now to watch an entire 30 minute TV show without turning the set off during commercials
she should give PBS a try