On Jul 2, 7:46 pm, Jason Moxham <ja...@njkfrudils.plus.com> wrote: > On Friday 02 July 2010 19:14:15 Cactus wrote: > > > > > On Jul 2, 7:04 pm, Jason Moxham <ja...@njkfrudils.plus.com> wrote: > > > On Friday 02 July 2010 18:46:33 Cactus wrote: > > > > On Jul 2, 5:00 pm, Jason Moxham <ja...@njkfrudils.plus.com> wrote: > > > > > On Friday 02 July 2010 14:52:14 Jason Moxham wrote: > > > > > > On Monday 28 June 2010 13:14:48 Jason Moxham wrote: > > > > > > > Hi > > > > > > > > Now we have removed the old cpu's here are some operating systems > > > > > > > I propose we remove explicit support for > > > > > > > > IRIX for mips > > > > > > > OSF/TRU64 for alpha > > > > > > > SunOS <=version 4 (version 5 is called solaris ie on > > > > > > > fulvia/mark.skynet) DJGPP dos > > > > > > > OS2 > > > > > > > Unicos cray's unix > > > > > > > pw32 posix on win32 > > > > > > > > Comments? > > > > > > > > Jason > > > > > > > I have removed all explicit support for the OS'es > > > > > > pw32 > > > > > > unicos > > > > > > os2 > > > > > > djgpp > > > > > > osf/tru64 > > > > > > > I have yet to do IRIX/SunOS as there quite a few simplifications > > > > > > that can be made. > > > > > > > I have removed the pre-build file fac_ui.h , the constants are now > > > > > > in the .c file , the program that generated them is in a new > > > > > > directory devel/ , which are files for the developers only , they > > > > > > will not appear in any mpir release.So we have yasm.diff(used for > > > > > > updating yasm) , setversion(used for changing version numbers) in > > > > > > there. > > > > > > > The windows build will need to reflect the fac_ui changes. > > > > > > > Jason > > > > > > I have removed the pre-build files mp_bases.h and fib_table.h and > > > > > incorporated them into gmp-impl.h , for windows the only thing to do > > > > > is that there is no need to generate them. > > > > > > Jason- Hide quoted text - > > > > > > - Show quoted text - > > > > > HI Jason > > > > > I have removed the three prebuild steps you have cahnged and now all > > > > the Windows builds fail with: > > > > > error LNK2001: unresolved external symbol __gmpn_bases > > > > > Where is this symbol supposed to be defined now? > > > > > Brian > > > > You are a bit premature , I have only removed half of them , the other > > > half still to do . I should do it by tomorrow . You might as well leave > > > what you have done for mo . I left it as autotools didn't like it when I > > > removed both , so I need a closer look at it. > > > > Thanks > > > Jason- Hide quoted text - > > > > - Show quoted text - > > > Thanks Jason - I will wait on your update before trying again. > > > Do you intend to remove the perfsqr.h prebuild step? > > Yep , they are all going :) , should finish the lot this weekend. >
All the pre-build stuff has been removed , and I have also removed the trunk/perfsqr.h file (that was "generated") and incorporated it into gmp-impl.h (like the other *.h generated files) I had to put a dummy file in , temporary , it's generated but not linked to the lib , this is to stop autotools complaining . I've put the new Slackware 13.1 on my "main" machine which has the new autotools , this conflicts with our current Makefile.am , so that needs to be sorted out , but it does allow us to run make check in parallel (or at least that's what is says on the tin) I got my main windows box back for the n'th time , but it's still broken , so windows stuff will have to wait I have removed the cray(vector machines) and VAX models form configure. IRIX/SunOS(<=4) OSes to be removed , bit by bit > > Brian > > > I assume I will still haveprebuiid steps > > I was thinking of ripping out the path selection code next. This is the code > in configure.in (about 1000 lines) which chooses which mpn asm code to include > and creates symbolic links. We could replace it with a table and a python? > script that we can share between linux and windows , I'm not too sure how this > would fit in with the project files? , but it could enable you to just do > just 1 > project with the cpu selection done in python. Actually I wont do it next. > > Two more cpu's I propose we drop support for > > cray > This is the cray vector machines which must be old supercomputers which by now > must be slower that most laptops , so I assume no one still uses them , some > have non-ieee floating point and 48bit ints , so there is quite a lot of > specific code for them. The latest cray machines are x86_86 (or slightly older > alpha's) > > vax > This is the 1970/80's minicomputer and according to wikipedia > > The VAX architecture was eventually superseded by RISC technology. In 1989 DEC > introduced a range of workstations and servers that ran Ultrix, the DECstation > and DECsystem respectively, based on processors that implemented the MIPS > architecture. In 1992 DEC introduced their own RISC instruction set > architecture, the Alpha AXP (later renamed Alpha), and their own Alpha-based > microprocessor, the DECchip 21064, a high performance 64-bit design capable of > running OpenVMS. > In August 2000, Compaq announced that the remaining VAX models would be > discontinued by the end of the year.[8] By 2005 all manufacturing of VAX > computers had ceased, but old systems remain in widespread use. > > Jason -- You received this message because you are subscribed to the Google Groups "mpir-devel" group. To post to this group, send email to mpir-de...@googlegroups.com. To unsubscribe from this group, send email to mpir-devel+unsubscr...@googlegroups.com. For more options, visit this group at http://groups.google.com/group/mpir-devel?hl=en.
