Hi I going to start on these autotools simplifications now , and hopefully the code is clean enough to finish it .
I appear to have my Windows box back alive and well , and after having some trouble with installation of Windows 64 (and 32) and MSVC , I should be able to give the Mingw64 (and 32) a go. Jason On Tuesday 27 July 2010 11:31:55 Jason wrote: > On Tuesday 27 July 2010 11:16:25 Bill Hart wrote: > > On 27 July 2010 11:09, Jason <ja...@njkfrudils.plus.com> wrote: > > > Hi > > > > > > Just thinking about the next bit of autotools simplifications , then > > > these bits are all interconnected in some way. > > > > > > Support for fat file systems(8+3 names) , ie we have a file > > > mpn/dive_1.c which gives us the function divexact_1 . We already dont > > > support fat file systems as we already have files with names longer > > > than 8+3 chars , so this is no great loss. So I propose to change the > > > file names to match the function names. > > > > This definitely sounds like a long overdue improvement. > > > > > Some files ie x86/aors_n.asm or mpn/generic/popham.c provide for two > > > functions , and the "decision" is made at compile time , I propose we > > > move the "decision" to "autotools" time. > > > > Do you mean have two symbolic links to the same file with different > > flags for compilation? > > Basically the same setup we have at the moment , but when we run autotools > , we run "our setup script" instead , which runs autotools AND "splits" > aors_n.asm into add_n.asm AND sub_n.asm , that way the build system > doesn't need the compilation FLAGS , ie the build system is now one > file=one function. The complication can still exist , but are confined to > our development machines , so we could write it in python(or whatever , > C?) > > > > There are lists of functions that have to be filled in various > > > Makefile.am 's , with the above changes we should be able to automate > > > it , and I think the Windows build could benefit from the code that > > > can list the files/functions. It would nice if this could handle the > > > function prototypes in the header files as well. > > > > This would be nice. > > There are of course files which can have multiple entry points , ie > mpn_add_n and mpn_add_nc , we would need to handle them , and I think > there are file which have a few functions in them (for tuning only?) . > Have to think about that.... > > > > I need to think about this some more , dont want to start it and get > > > half way through , and realize I should of done it a different way :) > > > > > > 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. -- 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.
