Nahh, that last bit will never work... Better idea, switch business methods with formal product delivery theories and economics theses, ruling them down to (relatively) solid science, which would be quite harder to emulate with something trivial. And in order to get a software patent, you must provide verbatim source code, possibly with a full language specification, and documentation. That would make software patents quite possible, but also - quite useless. And while we're at it - compiled binary is not a full fledged (!) derivative work of the source code, because it retains too little information from the original. So you can't actually license binaries in any form directly, only source code, and automatically give permission to use the binaries to any source code licensee. </crackpipe>
And I don't mind. What I do mind is that nobody is targeting me. Seriously, are there so little geeks out there that it doesn't make business sense, or is just another case of collective corporate idiotitis?
I've had it with semi-proprietary UNIX system integrators. Can't somebody slap PC-BSD with a ZFS root on a generic box, invest some in proper hardware support, and call it a day? I want nice UN*X system, but at a reasonable price, otherwise, I'd rather roll my own.
Oh, great... <flaimbait>Now, instead of needing an Exchange killer, we're gonna have to wrestle with turtlenecked fanbois with suits on top, and tear our asses making a freaking corporate-Itunes-for-the-IPhone replacement. </flamebait> Seriously, it's outta the frying pan and into the fire for FLOSS. Good thing those production servers aren't going anywhere, but still...
You know the old drill... Fork and fix. In FF's case, make an addon. Or take it up with Mozilla/Google. It's a user interface bug, far as I'm concerned.
Reflection is a type of metaprogramming, so anonymous functions, with some compiler tweaks can yield it. But, yes, I agree, the C++ standard should have focused on provisions for language features, and not features per se. And frankly, I don't see any modern programming paradigm that can't be implemented via pointers and anonymous functions. So why weren't they implemented? That isn't the C way. The C way to OOP is to write your own object system in 15 lines without noticing.
Perl? Too slow. There is already POSIX, all you have to do is setup some polymorphic code with built in workarounds for not so POSIX-compliant systems (Linux, Windows, I'm looking at you), and you get the same thing as the Perl exploit you described, only faster, and with more geek points.;)
Unless we make a ebuild2port* converter, and tweak pkgsrc to be able to build cross-distribution RPMs and DEBs.
* - An ebuild is a Gentoo Portage package. A port is a FBSD Ports Tree package. A pkg is a NBSD pkgsrc package. There does exist a port2pkg converter already, so I thought that as long as we're a it, we might as well give FBSD a hand as well, and not just NBSD.
Though I really think that the community needs such a license in order to effectively (under legal threat, that is) prevent scope/feature creep and fragmentation. Take a good look at those licenses in Wikipedia, and you'll understand better. That is, if you haven't already.
Sometimes I dream we used something in between CDDL, QPL, SISL, LGPL v3, and the LaTeX license. No more Affero variants for networked apps, just better wording. You have the right to distribute the app with your modifications only if it matches the original project specification, and/or the official standard(s), otherwise, it's back to tarballs and patches for you and your users. Good separation of concerns, no more excessive viral licensing, but making damn sure everybody is contributing back (on a per-[file/class/object/aspect/function/*] basis). Damn it, RMS, get it right for a change!
Slashdot Mirror
Somebody fork Slashcode, add NNTP and SMTP export options, and fix the freaking mod system already!
Nahh, that last bit will never work...
Better idea, switch business methods with formal product delivery theories and economics theses, ruling them down to (relatively) solid science, which would be quite harder to emulate with something trivial. And in order to get a software patent, you must provide verbatim source code, possibly with a full language specification, and documentation. That would make software patents quite possible, but also - quite useless.
And while we're at it - compiled binary is not a full fledged (!) derivative work of the source code, because it retains too little information from the original. So you can't actually license binaries in any form directly, only source code, and automatically give permission to use the binaries to any source code licensee. </crackpipe>
And I don't mind. What I do mind is that nobody is targeting me.
Seriously, are there so little geeks out there that it doesn't make business sense, or is just another case of collective corporate idiotitis?
Let me introduce you to the concept of Live distributions. They run off removable media. Just saying.
I've had it with semi-proprietary UNIX system integrators. Can't somebody slap PC-BSD with a ZFS root on a generic box, invest some in proper hardware support, and call it a day? I want nice UN*X system, but at a reasonable price, otherwise, I'd rather roll my own.
Oh, great...
<flaimbait>Now, instead of needing an Exchange killer, we're gonna have to wrestle with turtlenecked fanbois with suits on top, and tear our asses making a freaking corporate-Itunes-for-the-IPhone replacement. </flamebait>
Seriously, it's outta the frying pan and into the fire for FLOSS. Good thing those production servers aren't going anywhere, but still...
Doesn't AIX pull the same binary conf file crappola?
Department of redundancy department much?
HTTP is a protocol. How can it have bad performance? Overhead? Seriously, I'd like to know.
You know the old drill... Fork and fix. In FF's case, make an addon. Or take it up with Mozilla/Google. It's a user interface bug, far as I'm concerned.
Because they aren't pure functional ones. And they "support" procedural programming, in the sense that they use procedural semantics.
Q:What will the language of modern scientific computing be in 20 years?
A:I don't know, but it will be called Fortran.
By your logic everybody should be coding in Ada...
Not a bad idea actually...
I would like to subscribe to your newsletter.
Reflection is a type of metaprogramming, so anonymous functions, with some compiler tweaks can yield it. But, yes, I agree, the C++ standard should have focused on provisions for language features, and not features per se. And frankly, I don't see any modern programming paradigm that can't be implemented via pointers and anonymous functions. So why weren't they implemented? That isn't the C way. The C way to OOP is to write your own object system in 15 lines without noticing.
Perl? Too slow. There is already POSIX, all you have to do is setup some polymorphic code with built in workarounds for not so POSIX-compliant systems (Linux, Windows, I'm looking at you), and you get the same thing as the Perl exploit you described, only faster, and with more geek points. ;)
Somebody reinvented the Netscape Public License, apparently.
Just wanted to chip in this AOL thread and say another thanks to Marcus Aurelius.
Buddhism isn't a religion, it's a life philosophy.
IOW, we need a science called ethics, and people willing to study it at least in the basics...
Multiple pointers, multiple screens.
=Smidge
Unless we make a ebuild2port* converter, and tweak pkgsrc to be able to build cross-distribution RPMs and DEBs.
* - An ebuild is a Gentoo Portage package. A port is a FBSD Ports Tree package. A pkg is a NBSD pkgsrc package. There does exist a port2pkg converter already, so I thought that as long as we're a it, we might as well give FBSD a hand as well, and not just NBSD.
Though I really think that the community needs such a license in order to effectively (under legal threat, that is) prevent scope/feature creep and fragmentation. Take a good look at those licenses in Wikipedia, and you'll understand better. That is, if you haven't already.
GNOME ain't a GNU project...~
Now think aobut the letter G for a sec...
Can't you wrap up the drivers with some source glue and inline the binaries, compile, and run all the drivers that you need in one big one?
Sometimes I dream we used something in between CDDL, QPL, SISL, LGPL v3, and the LaTeX license. No more Affero variants for networked apps, just better wording. You have the right to distribute the app with your modifications only if it matches the original project specification, and/or the official standard(s), otherwise, it's back to tarballs and patches for you and your users. Good separation of concerns, no more excessive viral licensing, but making damn sure everybody is contributing back (on a per-[file/class/object/aspect/function/*] basis).
Damn it, RMS, get it right for a change!
No, the apps can, in theory...
There is only one that I heard could actually do it, but keep in mind that it Eventually Mallocs All Core Storage.