Tables are but a visual, partial, inacurate representation of relations. Tables aren't relational.
A DBMS that implements tables as primary objects instead of as simple representations of relations isn't relational: case in point, SQL and therefore FileMaker.
>
when you change some behaviour in a component *and* you change its interface, too, you have more options to deal with incompatibilities than when you change the behaviour and keep the interface
Who cares?
Sometimes you have to change the interface to get the best behaviour. And everyone wants binary-only modules to go away. So good riddance.
I mean, real Literature with an uppercase A, not only documentation.
Two interesting books about what if all your world was virtual:
A masterwork is Charles Williams' Descent into Hell. Here a historian gets, by special providence of a kind of a genie, to recreate his own beloved as he wishes instead of disputing her with a younger, more daring rival. By doing so he ends up cutting himself from all reality, thus effectively becoming irrecuperably mad - and also committing himself to Hell too in the process.
A lighter, more to the point, but also even more overtly Christian book to the point of being a bit doctrinaire, is CS Lewis's The Great Divorce. Here, Hell is just what happens when everyone can have whatever he wants; soon enough people discover they can't live near other people's ideas of what the world should be, and the result is that everyone but the newly arrived lives in perfect isolation at ever-crescent distances from one another.
When I read Namesys' papers some time ago, it struck me that they seemed to aim to integrate a simple hierarchical DBMS at the kernel disguised as a filesystem.
Now if that's correct, we should instead just go for a much simpler, better-performing RDBMS.
MS software runs on a too different platform, namely MS Win32; and embodies too bad development and administration practices. You have been improving, but be it the marketing smokescreen or whatever, sometimes it seems you only improve points without ever seeing the picture.
IMNSHO you really missed the boat when you scuttled the planned migration from MS DOS 2.11 or some successor of it to MS Xenix, deciding instead to keep patching MS DOS until it became real-mode MS DOS^H^H^H^H^H^H^H^H^H^H^H^H^H^H^H^H^H^HMS OS/2^H^H^H^H^H^H^HMS WNT^H^H^H^H^H^HMS.Net, whatever.
Now what you can do? Even if MS.Net is relatively good, the platform it runs on is just too complex. You could possibly migrate to a free software kernel like Linux, a BSD or even the GNU Hurd, but I doubt it. Making MS.Net a standard could help, but it would need to come with assurances of no aggression with patents or whatever.
MS SQL Server should not only strive for but indeed achieve ISO SQL compliance at least on par with IBM DB2.
Just as good or better would be to freely publish specs of protocols, interfaces and file formats. That would in effect preclude nearly all the antitrust nastiness, and create a level playing field with free software. It would, as a huge bonus, build the bridge with free software, as in enabling you to use more of it - apart from hoarding the BSD IP stack into MS Windows and using the GNU toolchain in MS UIS - and enabling the free software community in using more of your tools and platform.
Now as we are talking about this, why not go to the root of the problem?
One of the main reasons there are so many standards (and non-standards like your proprietary interfaces, protocols and formats) is complexity, and one of the main causes of complexity is lack of power deriving from poor fundamentals. Look at Lisp: one of the three oldest programming languages, it is also one of the simplest and at the same time more powerful. Even C and POSIX, bad as they are, have such longevity compared to MS platforms due to their inherent power.
One way here would be to collaborate with the GNU Hurd, that tries to stabilish a bridge between the POSIX (and why not MS Win32 and.Net) world with the functional programming world.
But an even better way would be relational data.
MS SQL Server may be nice enough as far as current SQL products go, bar IBM DB2 and PostgreSQL, and MS.Net has a real nice concept with its common type system. But you could both make MS.Net and an RDBMS successor to MS SQL Server much better if you made a sane type system the base of your whole platform, and integrated a D language-accessible TRDBMS compliant with Date's and Darwen's _The Third Manifesto_ at the core. The icing on the cake would be functional programming as a preferred systems programming language, but perhaps that'd be asking too much.
And what would that have to do with free software? Well, it would be such a leapfrog on systems development and deployment that it would stabilish whole new standards. Make them real standards - just as IBM did with ISO SQL -, and you won't really need to give much away to be a good member of the global village.
>
Id like to see a more rigid Linux Kernel API (so a module doesnt have to be recompiled between kernel maintenace versions), and the 'grey' area of binary modules sorted out as well. I dont think it will happen.
Why not?
As for a stable Linux modules API, it will happen once Linux is mature enough. Perhaps the whole monolithic kernel with procedural programming will prevent this, but barring wholesale migration to GNU Hurd with its microkernel architecture and hoped-for functional systems programming the alternative is a substandard API.
Given that binary modules are considered bad, it is a low priority goal.
it seems like a dead-end choice to buy yet another Alpha-based machine
Only that (1) Alpha still has more than good enough performance, (2) you stick to what you already have working, (3) competitors don't have yet a compelling story on the viability of their RISC offerings, (4) going Intel feels like downgrading, (5) HP's migration proposals are still ridiculous, because (a) there is no good substitute for Digital Unix yet, HP-UX being much inferior, (b) no one believes in Itanium.
>
The definition of "open" in IT is: "open as in ``open your wallet".
So you are unaware that open systems (as in Unix) are what has driven down the (formerly expensive) prices of proprietary systems (then IBM)?
Take open systems from us (as in, let each current POSIX system diverge enough) and you will see MS, VMS and IBM prices hiking even higher than currently.
>
the source code to Solaris is open
So try modifying and redistributing it to see how Sun likes it.
>
"open standard" is a euphemism for a protectionist practice where companies get together and try to protect their market against outsiders
You have it in reverse. Companies (and users, including bigcorps and govs) banded together to break open the market from single-company's 'de facto standards', such as IBM, Microsoft and Digital had and still have.
>
What we need is a real open standard, where only stuff is permitted in that is not covered by patents, can be freely used by everyone without any licensing claim by anyone, and where there exists a free software reference implemention.
Do you mean like POSIX, the Internet and other small, irrelevant stuff?
>
Do you have anything other than opinion backing up the claim you just made?
Yep. Just Google around for the Mac OS X kernel and its creator, plus perhaps Mach, BSD, kernel, microkernel. I could give you the guy's name and the name of the merged entity of BSD plus Mach, but I'm too lazy now.
>
it's my understanding that Darwin/MacOS X is a BSD personality layered on top of Mach...
I thought that too. But the 'layering' is actually a commingling.
Slashdot Mirror
Yeah, but...
...then you loose all kinds of data entry integrity checks, like zip code format per country, lists of valid states, etc.
It is only for English-speaking North American countries for now.
Lack of time to expand, or just a revenge for Orkut?
Not as well as NX.
Perhaps situation will change when everyone is doing vectorial graphics with Cairo sent differentially with XDamage, but for now NX is the way to go.
It works.
And BTW, it works with anything X11 too.
Tables are but a visual, partial, inacurate representation of relations. Tables aren't relational. A DBMS that implements tables as primary objects instead of as simple representations of relations isn't relational: case in point, SQL and therefore FileMaker.
I take it to mean 'partially ISO SQL compliant'. SQL isn't relational. The ISO standard doesn't even use the 'relational' word.
SCSI, SATA or what?
Who cares?
Sometimes you have to change the interface to get the best behaviour. And everyone wants binary-only modules to go away. So good riddance.
I mean, real Literature with an uppercase A, not only documentation.
Two interesting books about what if all your world was virtual:
A masterwork is Charles Williams' Descent into Hell. Here a historian gets, by special providence of a kind of a genie, to recreate his own beloved as he wishes instead of disputing her with a younger, more daring rival. By doing so he ends up cutting himself from all reality, thus effectively becoming irrecuperably mad - and also committing himself to Hell too in the process.
A lighter, more to the point, but also even more overtly Christian book to the point of being a bit doctrinaire, is CS Lewis's The Great Divorce. Here, Hell is just what happens when everyone can have whatever he wants; soon enough people discover they can't live near other people's ideas of what the world should be, and the result is that everyone but the newly arrived lives in perfect isolation at ever-crescent distances from one another.
Que pena... pensei que pudesse ser algo mais interessante.
Didn't get your joke.
When I read Namesys' papers some time ago, it struck me that they seemed to aim to integrate a simple hierarchical DBMS at the kernel disguised as a filesystem.
Now if that's correct, we should instead just go for a much simpler, better-performing RDBMS.
Does this thing has mechanical springs à la IBM Model M?
If not, and if I can't get a Brazilian layout, I have no use for it.
Just posted at his blog:
.Net, whatever.
.Net is relatively good, the platform it runs on is just too complex. You could possibly migrate to a free software kernel like Linux, a BSD or even the GNU Hurd, but I doubt it. Making MS .Net a standard could help, but it would need to come with assurances of no aggression with patents or whatever.
.Net) world with the functional programming world.
.Net has a real nice concept with its common type system. But you could both make MS .Net and an RDBMS successor to MS SQL Server much better if you made a sane type system the base of your whole platform, and integrated a D language-accessible TRDBMS compliant with Date's and Darwen's _The Third Manifesto_ at the core. The icing on the cake would be functional programming as a preferred systems programming language, but perhaps that'd be asking too much.
Standards and fundaments.
MS software runs on a too different platform, namely MS Win32; and embodies too bad development and administration practices. You have been improving, but be it the marketing smokescreen or whatever, sometimes it seems you only improve points without ever seeing the picture.
IMNSHO you really missed the boat when you scuttled the planned migration from MS DOS 2.11 or some successor of it to MS Xenix, deciding instead to keep patching MS DOS until it became real-mode MS DOS^H^H^H^H^H^H^H^H^H^H^H^H^H^H^H^H^H^HMS OS/2^H^H^H^H^H^H^HMS WNT^H^H^H^H^H^HMS
Now what you can do? Even if MS
MS SQL Server should not only strive for but indeed achieve ISO SQL compliance at least on par with IBM DB2.
Just as good or better would be to freely publish specs of protocols, interfaces and file formats. That would in effect preclude nearly all the antitrust nastiness, and create a level playing field with free software. It would, as a huge bonus, build the bridge with free software, as in enabling you to use more of it - apart from hoarding the BSD IP stack into MS Windows and using the GNU toolchain in MS UIS - and enabling the free software community in using more of your tools and platform.
Now as we are talking about this, why not go to the root of the problem?
One of the main reasons there are so many standards (and non-standards like your proprietary interfaces, protocols and formats) is complexity, and one of the main causes of complexity is lack of power deriving from poor fundamentals. Look at Lisp: one of the three oldest programming languages, it is also one of the simplest and at the same time more powerful. Even C and POSIX, bad as they are, have such longevity compared to MS platforms due to their inherent power.
One way here would be to collaborate with the GNU Hurd, that tries to stabilish a bridge between the POSIX (and why not MS Win32 and
But an even better way would be relational data.
MS SQL Server may be nice enough as far as current SQL products go, bar IBM DB2 and PostgreSQL, and MS
And what would that have to do with free software? Well, it would be such a leapfrog on systems development and deployment that it would stabilish whole new standards. Make them real standards - just as IBM did with ISO SQL -, and you won't really need to give much away to be a good member of the global village.
On one hand, MS keeps playing hard and dirty, won't backtrack, won't make amends, won't ask for excuses.
On the other, they developers want to play nice.
A house divided can't stand.
They have the source code for lots of COM classes that do that, so the need for docs is not that pressing.
Why not?
As for a stable Linux modules API, it will happen once Linux is mature enough. Perhaps the whole monolithic kernel with procedural programming will prevent this, but barring wholesale migration to GNU Hurd with its microkernel architecture and hoped-for functional systems programming the alternative is a substandard API.
Given that binary modules are considered bad, it is a low priority goal.
Non-sequitur.
To the contrary, the number of inactive projects is proof that the proposed natural selection method works at least partially.
Can you spell 'proprietary'?
Only that (1) Alpha still has more than good enough performance, (2) you stick to what you already have working, (3) competitors don't have yet a compelling story on the viability of their RISC offerings, (4) going Intel feels like downgrading, (5) HP's migration proposals are still ridiculous, because (a) there is no good substitute for Digital Unix yet, HP-UX being much inferior, (b) no one believes in Itanium.
Exactly my question!
Not even that, the commingling of BSD and Mach is obviously fatter and slower than simple BSD.
Hasta la vista, Windows! Or, Windows is the sickness. Linux is the cure.
No preferences of interpretation here, it is closed source but an open system conforming to (some) open standards.
So you are unaware that open systems (as in Unix) are what has driven down the (formerly expensive) prices of proprietary systems (then IBM)?
Take open systems from us (as in, let each current POSIX system diverge enough) and you will see MS, VMS and IBM prices hiking even higher than currently.
So try modifying and redistributing it to see how Sun likes it.
You have it in reverse. Companies (and users, including bigcorps and govs) banded together to break open the market from single-company's 'de facto standards', such as IBM, Microsoft and Digital had and still have.
Do you mean like POSIX, the Internet and other small, irrelevant stuff?
Yep. Just Google around for the Mac OS X kernel and its creator, plus perhaps Mach, BSD, kernel, microkernel. I could give you the guy's name and the name of the merged entity of BSD plus Mach, but I'm too lazy now.
I thought that too. But the 'layering' is actually a commingling.