I had an alpha/OSF for 5 years until March 2004; by then it was essentially the only such machine running Magma in the world with one exception being the machine they built Magma on for me. I think they were relieved when it went...
John On 9/24/07, mabshoff <[EMAIL PROTECTED]> wrote: > > On Sep 24, 5:22 pm, "William Stein" <[EMAIL PROTECTED]> wrote: > > On 9/24/07, Bill Hart <[EMAIL PROTECTED]> wrote: > > > > > OK, I hadn't read section 9 of the GPL. So now I agree with you that > > > Pari doesn't need to do anything. Excellent. > > > > :-) > > > > > So basically all the new lines of SAGE code would be a library, which > > > can be linked against other libraries, and would be v2 or later. > > > > Yep. > > > > > The SAGE tarball, that contains everything, including GMP, would be > > > licensed under v3. But everything in it will need to be "V2 or later" > > > or "v3 or later". At present only SINGULAR is a problem in this > > > regard. > > > > That's probably the case. Does anybody know anything about the > > plans for Singular (Martin? Michael?). > > > > I do not know. > > > > With regards to Magma, yes, they'll be forced to dynamically link > > > against GMP. I wouldn't put it past them to write their own library > > > and dump GMP though. Some of their core code clearly already uses > > > their own code. For example they use different code for FFT > > > multiplication of integers and clearly have a different algorithm for > > > integer division. They only support a few target platforms, so this is > > > not that hard for them to manage. > > > > Magma supports more target platforms than any other commercial > > mathematical software and more targets than Sage. If you go here: > > https://magma.maths.usyd.edu.au/magma/export/ > > you'll see they support all of: > > > > Alpha (Linux) > > Have access, but who cares about Linux ob Alpha? This should be fairly > easy, but after the Alphacide by Intel few alpha clusters should be > left. > > > Alpha (OSF/Tru64) > > Have access, but who cares about OSF on Alpha ? Shouldn't be too hard > assuming one builds his/her own toolchain. Chances are that if you own > an alpha that by now you have switched that box over to Linux - if you > need to ask why I just recommend using OSF for an hour or two :) > > > AMD64/Intel64 (Linux) > > Yep. > > > AMD64/Intel64 (Solaris) > > Close to getting it to work, but the 32 bit version also is close to > work. > > > i386/PC (Linux) > > Yep. > > > i386/PC (Windows) > > Well, this is a tricky one, especially AMD64/Intel64 - which Magma > doesn't seem to support, but I consider this possible. This should > overall dwarf the rest of the potential Sage installations, if you > look at high schools and not the university scene. But even in the > university scene there is a surprising amount of Windows around in my > experience. Getting at least Cygwin to (mostly) work again isn't that > hard, but it seems that nobody has the time to do so and I rather work > on Solaris than Windows. > > > IBM PowerPC64 (AIX) > > Nope, lack access, but not a very high priority in my opinion. > > > IBM PowerPC64 (Linux) > > Well, 32 bit runs and 64 bit isn't very difficult to do, just need > something faster than a PS3 to build it. We also need to do something > about multilib configurations for Sage in general. > > > Intel IA64 (Linux) > > Macintosh 32-bit Intel (OS X) > > Macintosh G4 (Mac OS X) > > Macintosh G5 (Mac OS X) > > Yep to all of the above. > > > Sparc (Solaris) > > Sparc64 (Solaris) > > Getting close. > > > > > It's actually pretty amazing. We should consider having similar > > support for Sage > > a challenge and priority. > > > > :) - well at least William has gotten my proposal about Sage/Windows/ > MSVC. > > > They won't jettison GMP. They also use several other LGPL (and soon > > to be LGPLv3) > > libraries, and I bet they'll just dynamically link them. After all, > > they already support > > that functionality (it's just not the default). > > > > > Possibly some other packages they use make use of GMP, and if *they* > > > go GPLv3 then likely Magma will just have to dynamically link to all > > > of those. Anyway, I think you are right. They will probably distribute > > > those libraries separately along with the LGPLv3 and just dynamically > > > link to them. That will give people the option of making drop in > > > replacements for them, according to the terms of the LGPLv3. > > > > Yep. That's what Maple already does too. > > > > It could be fun, because you could write a replacement to GMP that > > records all arithmetic operations that are done using GMP, then link > > Maple against it and see what Maple is doing... :-) > > > > Hehe, I would consider it next to impossible to catch up with GMP in > the short term. Pretty much any other arbitrary precision library out > there has folded. While it is certainly true that working with some of > the GMP people can be hard it seems to be much easier to carry a > couple patches. In regards to remedy the Core Duo situation I can > probably say something about that at SD5. > > > > Bill. > > <SNIP> > > Cheers, > > Michael > > > > > -- John Cremona --~--~---------~--~----~------------~-------~--~----~ To post to this group, send email to sage-devel@googlegroups.com To unsubscribe from this group, send email to [EMAIL PROTECTED] For more options, visit this group at http://groups.google.com/group/sage-devel URLs: http://sage.scipy.org/sage/ and http://modular.math.washington.edu/sage/ -~----------~----~----~----~------~----~------~--~---
