On 4 January 2011 00:22, Volker Braun <vbraun.n...@gmail.com> wrote: > I don't understand your point. If you have a 64-bit kernel but 32-bit > userspace then you can't just build a 64-bit Sage. Despite our worst > attempts we are still not shipping with glibc, ncurses, zlib, freetype, ... > My system, on the other hand, is multilib and I think I have all necessary > libraries to build both 64-bit and 32-bit Sage. I never tried the latter > because there is no SAGE32 nor any good reason to do so. > The manual says: "The SAGE64 variable is mainly of use is on OS X (pre > 10.6), Solaris and OpenSolaris".
There are some instances where building 32-bit executables is preferable on a 64-bit system. The main reason with Solaris is undoubtedly to maintain backwards compatibility. But it does mean the pointers are 32-bits, so more can fit in the cache. In some cases it can give an improvement in speed. At one point Mathematica's kernel on Solaris was 64-bit, but the front end was 32-bit. This was simply because Wolfram Research see no advantage, but some disadvantages in making the front end 64-bit. I don't know if the front end is 64-bit now, and can't be bothered to check. But it would not surprise me if it was still 64-bit. There are a couple of issues with R that mean building a 64-bit version of Sage on Solaris x86 is currently impossible. There are issues with several parts of Sage which mean that whilst Sage will build 64-bit on Solaris SPARC, it crashes very easily. Even before we get to the Sage prompt: sage: Sage has already leaked memory. That wont cause a crashes I see, but there are certainly some bugs that should really be found. There are some parts of Sage which will never benefit from being 64-bit, but it's less hassle to just make it all 64-bit. Despite Sage was supposedly 64-bit on OS X years ago, in fact parts of it would still have build 32-bit up until fairly recently. I would not be surprised if some of the optional or experimental Sage packages build 32-bit on early OS X systems even now. Dave -- To post to this group, send an email to sage-devel@googlegroups.com To unsubscribe from this group, send an email to sage-devel+unsubscr...@googlegroups.com For more options, visit this group at http://groups.google.com/group/sage-devel URL: http://www.sagemath.org
