>
these people just don't seem to be attracted to the open source community in the same way as programmers
I'd say it is just about timings and critical masses. At the moment, there isn't a designer critical mass in the free software movement, simply because Mac OS X is the designer platform and us geeks still need to catch up on apps funcionality.
It cuts both ways, as in Mac OS X doesn't have a critical mass of developers so there isn't much native free software ports for it.
Anyway I still maintain that to judge on such a general issue you still have to check Gnome 2, KDE 3.
>
It's that most free software I've tried on my Mac isn't aesthetically pleasing.
You missed the point. Software not conforming to the platform is, by definition, not sthetical. You should either evaluate native Mac OS X software, or evaluate say Gnome or KDE.
While Gnome, for instance, ain't as gorgeous as Mac OS X, it is far superior to That Other Proprietary Platform which actually regressed in pleasantness in its XP version.
>
the menubar in OS X is at the top of the screen because it's convenient there
So it is in most implementations of Gnome, and in those where it isn't by default (Red Hat's Bluecurve) it can easily be put there.
Not that I do not miss the Mac's global menu.
>
I think you may be underestimating KHTML's standards compliance nowadays.
I hope so.
>
Not to turn this into a KHTML vs Gecko flamewar.:-)
Not at all, I actually loathe XPCOM. It is just that I question the KHTML decision's expediency and wisdom.
I do hope that someday GtkHTML gets to KHTML levels of completeness, for I loathe C++ as well, and prefer Objective C.
>
None of the UI metaphors are consistent with what other Windows programs do. And yet... people generally regard it (in my experience) as an incredibly intuitive
I think here I am more radical than you UI-wise. I loathe Apple's brushed metal.
>
I'm talking about open source Aqua-native applications like GNUMail.app, Camino, a dozen HTML editors I've forgotten the names of, Mactella, MPlayer OS X, and on and on.
GNUMail isn't actually a GNUStep app? I don't know, does it 'gets' Aqua interface under Mac OS X -- as it should? If so, it would make a point in the case for other GNUStep apps. But I feel even Apple is not emphasizing Cocoa as it merited.
Camino was a development version only, as I pointed out. The others I know nothing about.
Anyway, you have to take into account what I described before about there not existing an incentive to develop native versions of free software to a proprietary platform.
>
I happen to think OS X has an aesthetic that is more beautiful and elegant than any other platform's
It is, even if it fails to be at least equal to Mac OS 9 on all points.
My point is just that Gnome 2, its HIG and native apps are More Than Good Enough.
>
Gnome and KDE give me a choice between 10,000 mediocre UIs
Not at all. I am using exclusively Gnome native software, except for GNU Emacs (a port is on the way, and I am sthetically almost as pleased as when I used Mac OS X -- minus the performance, compatibility, software availability and licensing headaches, but then I used Mac OS X 10.0, got screwed by Apple in licensing and iTools, and I am from a Third World country where we just don't have enough dollars to burn in Steve Job's altar.
It does no good to compare development versions to released software. Your question is another, the lack of standards (X, and Gnome or KDE integration) in your platform. Better than "that other proprietary platform, but Gnome (or KDE) sure would give you a different experience.
>
Why wouldn't I just use Safari under these circumstances?
My "problem" (not really, my Macs run Debian) is why Apple used something other than Gecko? Yes, I know KHTML is ligther than Gecko, presumably because Qt is POSIX native while XPCOM is a port from that other platform, but how long will it take for Safari to reach the compatibility level and features of Gecko, and how lighter and faster will it be then, if at all?
>
I don't care what is and isn't "native"; at the end of the day, I just want programs with well designed, aesthetically pleasing interfaces.
That's what only native programs can give you. Consider the free software programs that have been created or ported to Mac OS a bonus, since you're on a proprietary, non-standard platform. You really can't ask for much.
It would be similar to me asking why Adobe Acrobat Reader doesn't really fit nicely with my gorgeous Gnome desktops. It's a port, that's why. I consider myself lucky to have it at all for the few cases where GGV doesn't cut it yet.
>
so for most well-designed programs it should be relatively easy to maintain a native Aqua port
That's quite beside the point. It is hard to justify even that easy maintenance work when it goes for a minority platform whose community isn't really keen on (software) freedom. It "pays" more to develop more, better programs for, say, Gnome.
Now if Apple would really get behing Cocoa as an open platform, say submitting it to standards groups and putting Aqua and its requirements under the GNU GPL, perhaps it would steal some (free software development) thunder from Gnome, KDE... not likely at all.
>
As for me switching to Gnome
Never intended to mean that. Only that you can't evaluate (Gnome|KDE) software from some other platform than (Gnome|KDE).
>
I don't know why it doesn't surprise me that someone with aesthetic sensibilities would disavow RMS's political philosophies
I'd say Esthetics and Morals are quite orthogonal, if not exhaustively. Witness Leni Riefstenhal (or how-dya-spell-it), Hitler's friend and superb cinemaker.
>
it's been a few months at least, and I'm not even sure which version of Gnome was installed at the labs here at school
Yeah, free software ain't mature yet. Gnome 2 is pretty recent, but 2.2 was again a big improvement. 2.4 and 2.6 aren't such big changes, basically stabilising for limit usage and polishing.
>
I'm mostly talking about the dozens of open source projects I've tried to use on my Mac, which have ALL been sorely lacking from a UI perspective
You're running non-native apps, because Aqua is proprietary. You should either go for native apps (there was a Mozilla port to Aqua, wasn't there?) or use Gnome apps on Gnome itself to have a better idea.
>
we don't know how MS is going to roll out updates.
Good point. Perhaps they will have to charge; providing an up-to-date virus database is expensive, and letting it rot may infuriate users. They will still be angry, but at least MS will be able to tell the infected ones they didn't bother to subscribe...
But no, MS users who tend to infuriate have voted with their feet already... incidentally, this proves 95% of people with IT buying decision power are actually sheep.
>
quite a lot of the cost of MS's products goes towards customer service.
No, it doesn't. Read the article, he says services and support aren't part of their package, and is left to 'the channel' -- perhaps he means it is sunk under the Mancha's waves...
>
Try calling up Debian and getting them to answer some questions.
In fact it is more agreeable to use Debian's mailing lists, newsgroups and IRC channels than MS support.
Not to mention one can do a contract for Debian support and get someone on the phone or whatever.
>
got any idea why they quit doing that in Win95 and beyond?
The question is, why they did that in the first place?
If it was because they reckognised MS-DOS was insecure, then they thought MS-W32 would be better, as they at least pretended to think.
If it was because they had competition from DR-DOS, that competition ceased.
Then the question becomes why they resumed the practice. It might be because they reckognise MS-W32 is insecure and won't get better for a while at least, or because of competition from POSIX secure systems. Since Everyone Knows Unix is dying, these POSIX systems must be BSDs and GNU.
Allowing non-relational access to the database. Allowing duplicates, and non-values.
>
How does SQL mix the logical and physical structures?
For example, views are non-updateable relations.
>
you can build a database in postgres or oracle that breaks the relational model (by allowing duplicate rows, for instance), but you don't have to create such a database in these DBMSses.
OK, here we agree. You can always define keys, always use SELECT DISTINCT. It is just that it is a pain, will perform poorly and be severily limitated as compared to relational systems.
It pertains to how the data is logicaly structured. Nothing at all to do with the language you use to interact with it.
The language must follow the logical structure, can't you see that?
OK then, you could create a DBMS without a proper language, just an API. But that is not the case in SQL, nor in D&D's D.
>
Many RDBMSses allow you to use langauges other than SQL to interact with the data
First, these aren't RDBMSs you are talking about, but SQL DBMSs. Because they don't support any relational languages, they don't support relational databases (that's what you mean by 'logical structures'), and even if they did, SQL access would break Codd's 12 rules.
BTW, Alphora Dataphor do is a RDBMS (they call it 'virtual' for they don't yet have a storage engine), and it does have something like SQL. But to preserve the database sanity, they had to severily change SQL, creating a sanitised 'RealSQL'.
Second, these other languages do have to conform to SQL's limitations and stupidities, inclusive because the database is defined in non-relational SQL, and because SQL does mixes the logical and physical structures.
>
you can create queries on your schema in any of these RDBMSes in relational algebra.
Do that. Seriously, try it. Now how will that fare with databases defined with duplicates, NULLs, without real user-defined types etc?
>
there isn't a database I've seen that fully implements the relational model.
At this list there are some. Not all of them are mature or fully relational, but all are better than SQL was at their own time. Some are available, and at least one used in production systems.
>
the title of the Slashdot article and the linked Firebird project page both proclaim that it is relational.
That was my point, that they are in error.
That such a fundamental error is propagated even by DBMS implementors speaks truckloads about levels of knowledge in the industry.
>
And I could be wrong, but I thought SQL is specifically for relational databases.
It was meant to. Only that IBM suits have put programmers in charge of the project (System R), programmers who never got the relational model at all, and steadfastly refused to hear input from Mr Codd (the RM inventor) and others, so in the end Mr Codd left IBM in disgust. He got sick and had to stop evangelising the RM, and all these unrepentant SQL vendors never offered him anything for his invention.
>
There might be a reason why you're a DBA looking for a job.
Perhaps because people who have been trained instead of educated don't want to hire people who have a more solid knowledge than themselves?
Seriously, thanks, I've got a job now. Just forgot to update my.sig... will do RSN.
>
SQL is a query language, relational is a database design paradigm.
Stop short that paradigm crap. It is a theory, not a (ill-defined) paradigm.
Now, how would you feel if you had had a Math teacher who had taught you his own invention of a nifty algebra notation that didn't conform to the applicable Mathematical theory? Cheated, furious, or what?
>
Firebird is a relational database that uses SQL.
The Relational Model pertains not to the physical level (DBMS engines) but to the interaction level (data languages). SQL is not relational. Therefore, a SQL DBMS is not relational, no matter if its engine is capable of supporting a relational language too.
>
the exceptions are: tables without a primary key containing otherwise duplicate rows, and the results of multiple SELECT statements using the "UNION ALL" keyword
Anyway these are not sets, therefore not relations.
How'd ya feel if I invented an algebra where almost always gave a number as a result?
That is one thing you should never do. Never ever trust any web site as authoritative.
What could be good for you is to read the arguments critically, and think. Perhaps even get a book, either on you library or buying it -- Date's _An Introduction to Database Systems_ *is* the standard reference textbook in the area, so it's worth it.
BTW, all caps in a title is perfectly accepted practice.
>
why don't the RDBMS vendors change their products?
First, they are not RDBMS vendors. They are SQL (or quasi-SQL, as with Oracle and MySQL) DBMS (or quasi-DBMS, as with MySQL again) vendors.
Now, answers are easy. You might as well ask why it took Apple 20 years to have a real OS, or why MS don't have security at all. Do you want it spelled out? Ignorance, they are too rich, lazyness, and shoot-the-messenger, NIH mentalities.
>
why hasn't any newly developed RDBMS, ever since the advent of the relational model in the 1970's, chosen to implement the "proper" relational model
Just take a look at my own Relational DBMS implementations list at DMoz. Not all listed there are truly relational, but the best approximations at their own time. You can even get free downloads. None is SQL compliant.
If you had really read Stallman, you'd know he supports using non-copyleft for fast popularity gains when the free implementation is competing with proprietary ones. When there's only a free implementation, he (and I) supports copyleft in order to give free software a competitive advantage over proprietary one -- that's the explanation for ReadLine.
When the X Window System appeared, there was no competition whatsoever. Even now there is none that fits the bill. But there are proprietary derivations hoarded from the RI, and lots of proprietary components that makes free software life so much more difficult, esp. for RISC users. It is the same pattern that almost killed Unix BSD.
Am I? That was not the point, but that copyleft (in general) help prevent proprietary forking that has all but killed BSD, and therefore almost all Unix and X by extension.
Slashdot Mirror
Neither is Firebird, because it is SQL.
Yes indeed.
I'd say it is just about timings and critical masses. At the moment, there isn't a designer critical mass in the free software movement, simply because Mac OS X is the designer platform and us geeks still need to catch up on apps funcionality.
It cuts both ways, as in Mac OS X doesn't have a critical mass of developers so there isn't much native free software ports for it.
Anyway I still maintain that to judge on such a general issue you still have to check Gnome 2, KDE 3.
Exceptions. There simply isn't a critical mass.
Good opportunity to learn about the relational model and how SQL violates it!
You missed the point. Software not conforming to the platform is, by definition, not sthetical. You should either evaluate native Mac OS X software, or evaluate say Gnome or KDE.
While Gnome, for instance, ain't as gorgeous as Mac OS X, it is far superior to That Other Proprietary Platform which actually regressed in pleasantness in its XP version.
So it is in most implementations of Gnome, and in those where it isn't by default (Red Hat's Bluecurve) it can easily be put there.
Not that I do not miss the Mac's global menu.
I hope so.
Not at all, I actually loathe XPCOM. It is just that I question the KHTML decision's expediency and wisdom.
I do hope that someday GtkHTML gets to KHTML levels of completeness, for I loathe C++ as well, and prefer Objective C.
I think here I am more radical than you UI-wise. I loathe Apple's brushed metal.
GNUMail isn't actually a GNUStep app? I don't know, does it 'gets' Aqua interface under Mac OS X -- as it should? If so, it would make a point in the case for other GNUStep apps. But I feel even Apple is not emphasizing Cocoa as it merited.
Camino was a development version only, as I pointed out. The others I know nothing about.
Anyway, you have to take into account what I described before about there not existing an incentive to develop native versions of free software to a proprietary platform.
It is, even if it fails to be at least equal to Mac OS 9 on all points.
My point is just that Gnome 2, its HIG and native apps are More Than Good Enough.
Not at all. I am using exclusively Gnome native software, except for GNU Emacs (a port is on the way, and I am sthetically almost as pleased as when I used Mac OS X -- minus the performance, compatibility, software availability and licensing headaches, but then I used Mac OS X 10.0, got screwed by Apple in licensing and iTools, and I am from a Third World country where we just don't have enough dollars to burn in Steve Job's altar.
It does no good to compare development versions to released software. Your question is another, the lack of standards (X, and Gnome or KDE integration) in your platform. Better than "that other proprietary platform, but Gnome (or KDE) sure would give you a different experience.
My "problem" (not really, my Macs run Debian) is why Apple used something other than Gecko? Yes, I know KHTML is ligther than Gecko, presumably because Qt is POSIX native while XPCOM is a port from that other platform, but how long will it take for Safari to reach the compatibility level and features of Gecko, and how lighter and faster will it be then, if at all?
That's what only native programs can give you. Consider the free software programs that have been created or ported to Mac OS a bonus, since you're on a proprietary, non-standard platform. You really can't ask for much.
It would be similar to me asking why Adobe Acrobat Reader doesn't really fit nicely with my gorgeous Gnome desktops. It's a port, that's why. I consider myself lucky to have it at all for the few cases where GGV doesn't cut it yet.
That's quite beside the point. It is hard to justify even that easy maintenance work when it goes for a minority platform whose community isn't really keen on (software) freedom. It "pays" more to develop more, better programs for, say, Gnome.
Now if Apple would really get behing Cocoa as an open platform, say submitting it to standards groups and putting Aqua and its requirements under the GNU GPL, perhaps it would steal some (free software development) thunder from Gnome, KDE... not likely at all.
Never intended to mean that. Only that you can't evaluate (Gnome|KDE) software from some other platform than (Gnome|KDE).
I'd say Esthetics and Morals are quite orthogonal, if not exhaustively. Witness Leni Riefstenhal (or how-dya-spell-it), Hitler's friend and superb cinemaker.
Yeah, free software ain't mature yet. Gnome 2 is pretty recent, but 2.2 was again a big improvement. 2.4 and 2.6 aren't such big changes, basically stabilising for limit usage and polishing.
You're running non-native apps, because Aqua is proprietary. You should either go for native apps (there was a Mozilla port to Aqua, wasn't there?) or use Gnome apps on Gnome itself to have a better idea.
When was last time you used Gnome 2? From 2.2 it's actually quite simple and polished.
Good point. Perhaps they will have to charge; providing an up-to-date virus database is expensive, and letting it rot may infuriate users. They will still be angry, but at least MS will be able to tell the infected ones they didn't bother to subscribe...
But no, MS users who tend to infuriate have voted with their feet already... incidentally, this proves 95% of people with IT buying decision power are actually sheep.
There is no revenue, it's bundled. I guess you thought 'preserving revenue against competition', and that's what I was talking about.
No, it doesn't. Read the article, he says services and support aren't part of their package, and is left to 'the channel' -- perhaps he means it is sunk under the Mancha's waves...
In fact it is more agreeable to use Debian's mailing lists, newsgroups and IRC channels than MS support.
Not to mention one can do a contract for Debian support and get someone on the phone or whatever.
The question is, why they did that in the first place?
If it was because they reckognised MS-DOS was insecure, then they thought MS-W32 would be better, as they at least pretended to think.
If it was because they had competition from DR-DOS, that competition ceased.
Then the question becomes why they resumed the practice. It might be because they reckognise MS-W32 is insecure and won't get better for a while at least, or because of competition from POSIX secure systems. Since Everyone Knows Unix is dying, these POSIX systems must be BSDs and GNU.
Allowing non-relational access to the database. Allowing duplicates, and non-values.
For example, views are non-updateable relations.
OK, here we agree. You can always define keys, always use SELECT DISTINCT. It is just that it is a pain, will perform poorly and be severily limitated as compared to relational systems.
The language must follow the logical structure, can't you see that?
OK then, you could create a DBMS without a proper language, just an API. But that is not the case in SQL, nor in D&D's D.
First, these aren't RDBMSs you are talking about, but SQL DBMSs. Because they don't support any relational languages, they don't support relational databases (that's what you mean by 'logical structures'), and even if they did, SQL access would break Codd's 12 rules.
BTW, Alphora Dataphor do is a RDBMS (they call it 'virtual' for they don't yet have a storage engine), and it does have something like SQL. But to preserve the database sanity, they had to severily change SQL, creating a sanitised 'RealSQL'.
Second, these other languages do have to conform to SQL's limitations and stupidities, inclusive because the database is defined in non-relational SQL, and because SQL does mixes the logical and physical structures.
Do that. Seriously, try it. Now how will that fare with databases defined with duplicates, NULLs, without real user-defined types etc?
At this list there are some. Not all of them are mature or fully relational, but all are better than SQL was at their own time. Some are available, and at least one used in production systems.
That was my point, that they are in error.
That such a fundamental error is propagated even by DBMS implementors speaks truckloads about levels of knowledge in the industry.
It was meant to. Only that IBM suits have put programmers in charge of the project (System R), programmers who never got the relational model at all, and steadfastly refused to hear input from Mr Codd (the RM inventor) and others, so in the end Mr Codd left IBM in disgust. He got sick and had to stop evangelising the RM, and all these unrepentant SQL vendors never offered him anything for his invention.
Perhaps because people who have been trained instead of educated don't want to hire people who have a more solid knowledge than themselves?
Seriously, thanks, I've got a job now. Just forgot to update my .sig... will do RSN.
Stop short that paradigm crap. It is a theory, not a (ill-defined) paradigm.
Now, how would you feel if you had had a Math teacher who had taught you his own invention of a nifty algebra notation that didn't conform to the applicable Mathematical theory? Cheated, furious, or what?
The Relational Model pertains not to the physical level (DBMS engines) but to the interaction level (data languages). SQL is not relational. Therefore, a SQL DBMS is not relational, no matter if its engine is capable of supporting a relational language too.
Anyway these are not sets, therefore not relations.
How'd ya feel if I invented an algebra where almost always gave a number as a result?
That is one thing you should never do. Never ever trust any web site as authoritative.
What could be good for you is to read the arguments critically, and think. Perhaps even get a book, either on you library or buying it -- Date's _An Introduction to Database Systems_ *is* the standard reference textbook in the area, so it's worth it.
BTW, all caps in a title is perfectly accepted practice.
First, they are not RDBMS vendors. They are SQL (or quasi-SQL, as with Oracle and MySQL) DBMS (or quasi-DBMS, as with MySQL again) vendors.
Now, answers are easy. You might as well ask why it took Apple 20 years to have a real OS, or why MS don't have security at all. Do you want it spelled out? Ignorance, they are too rich, lazyness, and shoot-the-messenger, NIH mentalities.
Just take a look at my own Relational DBMS implementations list at DMoz. Not all listed there are truly relational, but the best approximations at their own time. You can even get free downloads. None is SQL compliant.
SQL is not relational. Its tables are not relations, because relations are sets, and sets don't contain duplicates.
There are several other reasons why SQL (and therefore derived products) aren't relational, check Database Debunkings for more info.
Firebird is SQL, not relational.
Why not? They were there before. They were the ones to be trampled upon by the great big dino.
If you had really read Stallman, you'd know he supports using non-copyleft for fast popularity gains when the free implementation is competing with proprietary ones. When there's only a free implementation, he (and I) supports copyleft in order to give free software a competitive advantage over proprietary one -- that's the explanation for ReadLine.
When the X Window System appeared, there was no competition whatsoever. Even now there is none that fits the bill. But there are proprietary derivations hoarded from the RI, and lots of proprietary components that makes free software life so much more difficult, esp. for RISC users. It is the same pattern that almost killed Unix BSD.
Am I? That was not the point, but that copyleft (in general) help prevent proprietary forking that has all but killed BSD, and therefore almost all Unix and X by extension.
For libraries, if you like, there's always LGPL.