Until there is a reasonable prototype, this has ALWAYS been the better approach. You have a small team that had the initial idea run things out for a while -- keep refining things as far as they can go. That will produce quicker progress for the immediate short term (i.e. getting them started). It is only later that the open source approach tends to become useful -- when useful ideas relating to the project, and bugs in implementation, get more obscure, thus requiring more eyes.
It was always said in the open-source whitepapers that successful open source projects always had reasonable code at the outset, not just a pretty idea.
John
You mean by 'cardinals' the set of natural numbers? There is no set of cardinals, in the case where you allow infinite cardinals (where by 'cardinal' I mean one of the 'cardinals', which is the proper-class of transitive sets well-ordered by inclusion).
'Trasitive' in this sense basically means that X is transitive if and only if for every subset Y of X, Y is also a member of X. 'Well-ordered by' means that every subset has a smallest element with respect to that ordering. John
In mathematics (or at least the kind I know of), 'countable' is a technical term meaning that the set can be placed into one-to-one correspondence with either the whole set of natural numbers, or with a finite subset of them (the former is said to be 'countably infinite', the latter 'finite'). The term 'uncountable' referes to the case when there are 'too many' of something to give each one a unique positive whole number (for example, the real numbers -- or any power set of an infinite set). John
I think, personally, that Stallman's use of the word 'forgiveness' is the sole problem.
He DID have a point.
Though I should add that it is a LEGAL point, and not a moral one (which everyone took it to be).
Now. The problem that he pointed out is a legal grey (gray in US speak) area -- and lest you not understand, lawyers on the other side LOVE grey areas.
The grey area is the fact that previous KDE software is technically not allowed to be linked with Qt in the way that it was. If KDE authors explicitly granted permission, then there would be no problem whatsoever -- but they didn't, possibly correcltly in terms of moral grounds, but in any case, they didn't.
The problem, logically speaking, is this. The KDE software that was (legally incorrectly) linked with Qt is not allowed to be distributed. Hence people who have worked on it -strictly-legally- have no rights to the software, and hence -strictly-legally- have no rights to distribute changes and changed software back, so the main KDE tree has not -strictly-legally- obtained rights to the software that it has merged into itself, EVEN THOUGH the authors all intend that to be the case.
Essentially, there is an assumed notion of permission, but no explicit lines drawn as to where this permission ends...
The problem with this is not that the free-software/open-source community are directly affected. Neither is it that the KDE project, on moral grounds, have done anything wrong (and this is where Stallman used the wrong words, by insinuating further wrongdoing where there was none). The problem is that software lawyers, acting for someone who has written in intentionally proprietary product, based upon KDE (which no-one in either camp wants to see released as proprietarty), could use that IMPLICIT notion of permission as a means to ALLOW THE PROPRIETARY PRODUCT THROUGH
Lest people not see it from that -- it gives those who wish to harm the GPL a powerful weapon to use in a test case in court. And the resulting precedent could be disasterous. This is what (I think) Stallman's worried about -- and given the nature of lawyers, probably not unjustly worried.
And that paragraph 9 is enforceable in the country of the author (which I don't think it is).
The question posed is about what happens if the code leaves the US to another country, the no-reverse-engineer clause becomes non-enforceable in that country, someone in that new country reverse engineers, writes specs, writes new code, and then the new specs and code make their way back to the US.
No US law has been broken (since the 'infringements' did not take place under US jurisdiction), so presumably this has the same effect as 'chinese walls'. The intellectual property in question does not belong to Verant. John
On the other hand, reusing a finely tuned skill (i.e. keyboard operation) in the context of a game such as Starcraft takes some serious beating.
Being able to look at a bunch of units, think what you want them to do, and have your finger simply hit the required key with the minimum of thinking, and no requirement for visual feedback during selection, makes the difference in a serious firefight.
I simply cannot see how a cumbersome controller can make up for that -- it cannot send more distinct signals to the computer, since you can physically hit more keys more quickly. It cannot give more precision when selecting and ordering -- pointing is one thing, but being able to quickly select multiple units and give them orders is a powerful ability: saving and recalling up to X (can't remember how many) different selections is crucial to managing the large numbers of units quickly and efficiently.
That said, with the lack of processor performance, I expect that the AI has been downgraded, so you'll stand a better chance of getting away with the input devices given. John
It wasn't so long ago that 3Dfx were suing nVidia for having two parallel texture processors (or something like that) -- this is merely nVidia giving some back. John
IIRC, Digital licensed an ARM core, deverloped it into StrongARM, had an argument with Intel, throwing a few lawyers around in the process, and ended up with Intel buying Digital's chip fabs, the StrongARM products, and a few other things besides.
Besides -- if we keep putting more and more up there, without clearing out the old stuff, then eventually we'll have some catastrophic disaster that'll take out an awfully large amount of satellites.
Bear in mind that (unless I'm mistaken), NASA is already thinking about the problem of how to get rid of space debris. Basically, the final act of ANY space project MUST be to clear up like this. Also, in the case of larger objects, where does the liability go if it comes out of orbit and levels somebody's house?? John
Qt wasn't the first to invent signals and slots -- and it wont be the last. Nor is it the best implementation of the concept. There is much ground to cover, and little to complain about if someone else wants to try out a new idea. John
If you want to program X, then poor font support comes with the territory. The freetype project are probably the ones to spearhead a decent improvement in this area -- so far as free software goes. John
Re:I actually have good things to say about Mozill
on
Mozilla M17 Is Out
·
· Score: 1
I woudln'd disagree about IE5. But what has to be said about the basic design for Mozilla is that it is designed to be a 'pervasive' web browser. It's small enough to get into every place that matters (or at least will be -- when it gets released proper), and has nearly all of the basic features required. The non-windows market is growing, and only time will tell if Mozilla is positioned correctly to take advantage of it. John
Or maybe just because 'a sledgehammer to crack a walnut' is what it takes to make any progress with an x86 design these days. Diminshing returns and all that... John
Basically, the Sledgehammer's aims are more for the short term. Intel have been desparate to get rid of the x86 for most if the IA's lifetime.
They designed it with a 3-4 year life-cycle in mind, then got themselves shot in the foot with it approaching a life-cycle time of more like ten times that!
The point of Intel's move into VLIW in the way that they have is to do to RISC what RISC did to CISC. Obviously, the persistance of CISC would seem to indicate that they are in for a rough ride, but the fact that RISC architecture has dominated so far as processor design innovation is concerned (note that most of the clever innovations on todays x86 chips are due to previous RISC designs).
p.s. to those who care as to where I may draw the lines bewteen RISC and CISC -- I draw them based upon the thinking behind them, which I take to be paramount. John
Hopefully, a lot of modern improvements in thinking in programming language design will be brought to bear in what picks up where Objective C left of. It, as a language, has many lessons to learn -- and the growing lack of its use gives it precisely the opportunity it needs to evolve and learn those lessons. John
So far as a generally accessible langauge is concerned, in ObjC it is far too easy to write
a program that can cause an unexpected core dump. Sure it makes many memory management problems easier in some respects -- but rarely does it make them no-brainers. And I'm afraid that that is what is required in this day and age -- unfortunately too many people do not have the time to bother with memory management, they just want to 'write their program and go...'.
TOM is an interesting, marginal, 'currently small-time' language that builds useful ideas upon the syntax of ObjC; but in truth, there is much yet to be discovered about how to write really good programming languages. So long as we don't forget ObjC, little will be lost if it ends up getting replaced. And even that seems unlikely given the stance of the GNUstep project. John
Is the PC a ball and chain? In general yes. It's architecture is the lowest common denominator of those available.
Is windows a ball and chain? Yes. It ties people to the PC architecture, and a whole lot of ancient cruft we'd rather be rid of.
What does Linux bring to this mess? Basically it gives you a short term loss (lack of immediately useable applications) for a powerful long term gain (lack of vendor lock-in -- i.e. not being heavily tied to one architecture and the problems that go with it --- in particular, it is easy to change). John
Most of the blame there lies with X. X encourages you to write your own... as a result, we have Motif, which was a 'standard' but was proprietary, closed, and unusable to both parties, Qt 1 and Qt 2's own home-grown solutions -- again unworkable from the GNOME camp's point of view, and finally GNOME -- which cannot really be made compatible with Qt with abandoning performance, speed, a useful API, etc. etc. completely.
This should clearly illustrate the need for policy to be BUILT INTO THE WINDOW SYSTEM AT DESIGN TIME -- not build time, not hack time, and certainly no later!! John
KOffice looks sweet but... For one thing, I could't insert an equation in the text at all. You can only insert (apparently at least) an object in a frame, and you can't have a frame dynamically resize itself, not move with a block of text.
Both of these are rather serious problems, which M$ Word had fixed very early on... John
The fatal mistake of mozilla was that they HAD the rendering engine (and still have). They tried to do too much with it. For example, someone's just knocked together a mozilla based browser with a simple fontend -- what most of us waht. Again, we don't mozilla to do our mail -- though it would be nice to have it to render html mail.
Basically, the mozilla project had what they needed over a year ago (i.e. the new rendering engine etc.) but have been faffing around with fancy bits and pieces that they wrongly thought essential for a first release. John
I'm running reiserfs with journaling on 2.2.13. That works fine (so fine that I've moved everything except root,home,archive partitions on my machine over to it). That said, reiserfs with journaling on 2.4.0test does NOT read them (i.e. refuses to mount).
This is a problem that I have, and is probably one that THEY know of, and is a good reason not to include reiserfs as part of the basic install.
Including the reiserfs patch (as part of an upgraded kernel source archive), and providing ext2-to-reiser utils (if and when they are written) would help things a good deal.
p.s. I haven't bothered to check/contact the reiserfs mailing list over this. This is not an excuse for me not having the problem 'fixed' -- 2.2.13 works fine for me. However it is a valid excuse for not having reiserfs in -their- kernel at this point.
p.p.s patience. I'd personally like to see LVM first, and reiserfs to wait in the wings until it is virtually (i.e. at least appears) designed for LVM. John
Apple can sue Cobalt -- NeXT designed the 'NeXT' cube well before Cobalt existed. Cobalt has been around too long for Apple/NeXT to claim a trademark over the cube design (they would have to have enforced sooner), though they have ample weight behind their side given that Apple has the rights to the original NeXT cube. John
Slashdot Mirror
Until there is a reasonable prototype, this has ALWAYS been the better approach. You have a small team that had the initial idea run things out for a while -- keep refining things as far as they can go. That will produce quicker progress for the immediate short term (i.e. getting them started). It is only later that the open source approach tends to become useful -- when useful ideas relating to the project, and bugs in implementation, get more obscure, thus requiring more eyes.
It was always said in the open-source whitepapers that successful open source projects always had reasonable code at the outset, not just a pretty idea.
John
You mean by 'cardinals' the set of natural numbers? There is no set of cardinals, in the case where you allow infinite cardinals (where by 'cardinal' I mean one of the 'cardinals', which is the proper-class of transitive sets well-ordered by inclusion).
'Trasitive' in this sense basically means that X is transitive if and only if for every subset Y of X, Y is also a member of X. 'Well-ordered by' means that every subset has a smallest element with respect to that ordering.
John
In mathematics (or at least the kind I know of), 'countable' is a technical term meaning that the set can be placed into one-to-one correspondence with either the whole set of natural numbers, or with a finite subset of them (the former is said to be 'countably infinite', the latter 'finite'). The term 'uncountable' referes to the case when there are 'too many' of something to give each one a unique positive whole number (for example, the real numbers -- or any power set of an infinite set).
John
I think, personally, that Stallman's use of the word 'forgiveness' is the sole problem.
He DID have a point. Though I should add that it is a LEGAL point, and not a moral one (which everyone took it to be).
Now. The problem that he pointed out is a legal grey (gray in US speak) area -- and lest you not understand, lawyers on the other side LOVE grey areas.
The grey area is the fact that previous KDE software is technically not allowed to be linked with Qt in the way that it was. If KDE authors explicitly granted permission, then there would be no problem whatsoever -- but they didn't, possibly correcltly in terms of moral grounds, but in any case, they didn't.
The problem, logically speaking, is this. The KDE software that was (legally incorrectly) linked with Qt is not allowed to be distributed. Hence people who have worked on it -strictly-legally- have no rights to the software, and hence -strictly-legally- have no rights to distribute changes and changed software back, so the main KDE tree has not -strictly-legally- obtained rights to the software that it has merged into itself, EVEN THOUGH the authors all intend that to be the case. Essentially, there is an assumed notion of permission, but no explicit lines drawn as to where this permission ends...
The problem with this is not that the free-software/open-source community are directly affected. Neither is it that the KDE project, on moral grounds, have done anything wrong (and this is where Stallman used the wrong words, by insinuating further wrongdoing where there was none). The problem is that software lawyers, acting for someone who has written in intentionally proprietary product, based upon KDE (which no-one in either camp wants to see released as proprietarty), could use that IMPLICIT notion of permission as a means to ALLOW THE PROPRIETARY PRODUCT THROUGH
Lest people not see it from that -- it gives those who wish to harm the GPL a powerful weapon to use in a test case in court. And the resulting precedent could be disasterous. This is what (I think) Stallman's worried about -- and given the nature of lawyers, probably not unjustly worried.
I just hope that somebody reads this
John
And that paragraph 9 is enforceable in the country of the author (which I don't think it is).
The question posed is about what happens if the code leaves the US to another country, the no-reverse-engineer clause becomes non-enforceable in that country, someone in that new country reverse engineers, writes specs, writes new code, and then the new specs and code make their way back to the US.
No US law has been broken (since the 'infringements' did not take place under US jurisdiction), so presumably this has the same effect as 'chinese walls'. The intellectual property in question does not belong to Verant.
John
On the other hand, reusing a finely tuned skill (i.e. keyboard operation) in the context of a game such as Starcraft takes some serious beating.
Being able to look at a bunch of units, think what you want them to do, and have your finger simply hit the required key with the minimum of thinking, and no requirement for visual feedback during selection, makes the difference in a serious firefight.
I simply cannot see how a cumbersome controller can make up for that -- it cannot send more distinct signals to the computer, since you can physically hit more keys more quickly. It cannot give more precision when selecting and ordering -- pointing is one thing, but being able to quickly select multiple units and give them orders is a powerful ability: saving and recalling up to X (can't remember how many) different selections is crucial to managing the large numbers of units quickly and efficiently.
That said, with the lack of processor performance, I expect that the AI has been downgraded, so you'll stand a better chance of getting away with the input devices given.
John
It wasn't so long ago that 3Dfx were suing nVidia for having two parallel texture processors (or something like that) -- this is merely nVidia giving some back.
John
IIRC, Digital licensed an ARM core, deverloped it into StrongARM, had an argument with Intel, throwing a few lawyers around in the process, and ended up with Intel buying Digital's chip fabs, the StrongARM products, and a few other things besides.
I may be wrong though.
John
FORTRAN, however, does predate emacs by quite a few years...
John
Besides -- if we keep putting more and more up there, without clearing out the old stuff, then eventually we'll have some catastrophic disaster that'll take out an awfully large amount of satellites.
Bear in mind that (unless I'm mistaken), NASA is already thinking about the problem of how to get rid of space debris. Basically, the final act of ANY space project MUST be to clear up like this. Also, in the case of larger objects, where does the liability go if it comes out of orbit and levels somebody's house??
John
Qt wasn't the first to invent signals and slots -- and it wont be the last. Nor is it the best implementation of the concept. There is much ground to cover, and little to complain about if someone else wants to try out a new idea.
John
If you want to program X, then poor font support comes with the territory. The freetype project are probably the ones to spearhead a decent improvement in this area -- so far as free software goes.
John
I woudln'd disagree about IE5. But what has to be said about the basic design for Mozilla is that it is designed to be a 'pervasive' web browser. It's small enough to get into every place that matters (or at least will be -- when it gets released proper), and has nearly all of the basic features required. The non-windows market is growing, and only time will tell if Mozilla is positioned correctly to take advantage of it.
John
Or maybe just because 'a sledgehammer to crack a walnut' is what it takes to make any progress with an x86 design these days. Diminshing returns and all that...
John
Basically, the Sledgehammer's aims are more for the short term. Intel have been desparate to get rid of the x86 for most if the IA's lifetime. They designed it with a 3-4 year life-cycle in mind, then got themselves shot in the foot with it approaching a life-cycle time of more like ten times that!
The point of Intel's move into VLIW in the way that they have is to do to RISC what RISC did to CISC. Obviously, the persistance of CISC would seem to indicate that they are in for a rough ride, but the fact that RISC architecture has dominated so far as processor design innovation is concerned (note that most of the clever innovations on todays x86 chips are due to previous RISC designs).
p.s. to those who care as to where I may draw the lines bewteen RISC and CISC -- I draw them based upon the thinking behind them, which I take to be paramount.
John
Hopefully, a lot of modern improvements in thinking in programming language design will be brought to bear in what picks up where Objective C left of. It, as a language, has many lessons to learn -- and the growing lack of its use gives it precisely the opportunity it needs to evolve and learn those lessons.
John
He who wields the knife -- never wears the crown.
John
Objective C has a lot for it, and a lot against.
So far as a generally accessible langauge is concerned, in ObjC it is far too easy to write a program that can cause an unexpected core dump. Sure it makes many memory management problems easier in some respects -- but rarely does it make them no-brainers. And I'm afraid that that is what is required in this day and age -- unfortunately too many people do not have the time to bother with memory management, they just want to 'write their program and go...'.
TOM is an interesting, marginal, 'currently small-time' language that builds useful ideas upon the syntax of ObjC; but in truth, there is much yet to be discovered about how to write really good programming languages. So long as we don't forget ObjC, little will be lost if it ends up getting replaced. And even that seems unlikely given the stance of the GNUstep project.
John
Is the PC a ball and chain? In general yes. It's architecture is the lowest common denominator of those available.
Is windows a ball and chain? Yes. It ties people to the PC architecture, and a whole lot of ancient cruft we'd rather be rid of.
What does Linux bring to this mess? Basically it gives you a short term loss (lack of immediately useable applications) for a powerful long term gain (lack of vendor lock-in -- i.e. not being heavily tied to one architecture and the problems that go with it --- in particular, it is easy to change).
John
Most of the blame there lies with X. X encourages you to write your own... as a result, we have Motif, which was a 'standard' but was proprietary, closed, and unusable to both parties, Qt 1 and Qt 2's own home-grown solutions -- again unworkable from the GNOME camp's point of view, and finally GNOME -- which cannot really be made compatible with Qt with abandoning performance, speed, a useful API, etc. etc. completely.
This should clearly illustrate the need for policy to be BUILT INTO THE WINDOW SYSTEM AT DESIGN TIME -- not build time, not hack time, and certainly no later!!
John
KOffice looks sweet but... For one thing, I could't insert an equation in the text at all. You can only insert (apparently at least) an object in a frame, and you can't have a frame dynamically resize itself, not move with a block of text.
Both of these are rather serious problems, which M$ Word had fixed very early on...
John
The fatal mistake of mozilla was that they HAD the rendering engine (and still have). They tried to do too much with it. For example, someone's just knocked together a mozilla based browser with a simple fontend -- what most of us waht. Again, we don't mozilla to do our mail -- though it would be nice to have it to render html mail.
Basically, the mozilla project had what they needed over a year ago (i.e. the new rendering engine etc.) but have been faffing around with fancy bits and pieces that they wrongly thought essential for a first release.
John
I'm running reiserfs with journaling on 2.2.13. That works fine (so fine that I've moved everything except root,home,archive partitions on my machine over to it). That said, reiserfs with journaling on 2.4.0test does NOT read them (i.e. refuses to mount).
This is a problem that I have, and is probably one that THEY know of, and is a good reason not to include reiserfs as part of the basic install.
Including the reiserfs patch (as part of an upgraded kernel source archive), and providing ext2-to-reiser utils (if and when they are written) would help things a good deal.
p.s. I haven't bothered to check/contact the reiserfs mailing list over this. This is not an excuse for me not having the problem 'fixed' -- 2.2.13 works fine for me. However it is a valid excuse for not having reiserfs in -their- kernel at this point.
p.p.s patience. I'd personally like to see LVM first, and reiserfs to wait in the wings until it is virtually (i.e. at least appears) designed for LVM.
John
Nobody's yet 'designed' a good OS, and nobody's designing them anymore... just a little scary.
John
Apple can sue Cobalt -- NeXT designed the 'NeXT' cube well before Cobalt existed. Cobalt has been around too long for Apple/NeXT to claim a trademark over the cube design (they would have to have enforced sooner), though they have ample weight behind their side given that Apple has the rights to the original NeXT cube.
John