On Thursday 22 July 2010 15:53:49 jason wrote: > I have tested the new autotools branch with our full mpir test > script , and the only issue over trunk is make check fails for yasm on > taurus and eno ONLY (releases do not run make check for yasm), this is > a minor thing and can be sorted out later. > In trunk I have managed to break the fat build on sextus(netburst) > ONLY , this is a bit of an oddity.
I've fixed that , a fat build requires asm functions , not plain C > > So I think I should push the new autotools into trunk. > > Jason > > On Jul 22, 1:30 am, Jason <ja...@njkfrudils.plus.com> wrote: > > On Wednesday 21 July 2010 23:21:11 jason wrote: > > > On Jun 2, 10:18 pm, Jason Moxham <ja...@njkfrudils.plus.com> wrote: > > > > Hi , here are some thoughts about what we should/could do for the > > > > next release. > > > > > > > > 1) Upgrade yasm to the latest (easy) > > > > > > > > 2) Upgrade gnu config to the latest ( dont know how difficult that is > > > > , but it could fix some niggles we have , and it might simplify our > > > > specialisations. > > > > > > > > 3) Upgrade to the latest autotools/libtool ( some distros are moving > > > > over to the latest 2.2 , we may need/want to do the same , again > > > > dont know what this involves) > > > > > > upgrading to the new autotools fixes the issues on my machine . see > > > branch new-autotools > > > It does mean that boxen can not be used for running autotools(ie make > > > dist) , eno however is upto date . We can still build mpir on all > > > machines (of course) but if you are going to run autotools , you need > > > the 2.2 (not 1.5) versions. > > > There is a minor problem that yasm fails make check on some machines , > > > and I will test(full test) it on the whole of skynet etc before > > > putting it into trunk. > > > > > > The update should enable a parallel make check , see item 8 below > > > > > > > 4) A few assembler functions to add > > > > > > > > 5) Move demo's out of the library onto the web page > > > > > > > > 6) Get rid of ancient cpus/compilers ( we will still work under C , > > > > if anyone cares) , this would simplify configure a bit , > > > > cray,pyramid, z8000,list,clipper,.... > > > > > > > > 7) Make configure run faster , I'm sure we can remove some of the > > > > tests , I can't believe they are still needed , and/or share the > > > > test results with yasm. > > > > > > > > 8) make make check parallel , it can be done. > > > > with the new autotools we can do a make check in parallel > > eg on eno > > > > without > > real 1m8.886s > > user 0m50.322s > > sys 0m6.525s > > > > with > > > > real 0m26.063s > > user 0m50.829s > > sys 0m7.377s > > > > the yasm test are not parallel so we dont show all the benefits , but for > > cygwin or solaris this will show a much better gain > > > > > > 9) Some of the changes we have made , have not been finished , finish > > > > them. > > > > > > > > 10) Split configure into two , ie standard and MPIR specific , should > > > > make the maintenance easier , this is fairly ambitious :) > > > > > > > > 11) drop support for building/running on FAT file systems ( ie file > > > > name 8.3 format) > > > > > > > > 12) simple command line build for windows ( not dependant of vcproj > > > > files) > > > > > > > > 13) fix some known bugs > > > > > > > > 14) When we update stuff , there are many places where you have to > > > > fill in the same info , make it automatic (autotools can do this , > > > > it's just not been set up that way) > > > > > > > > Some of these are trickier than others , but my aim is to simplify > > > > the system (the non-computational parts of it) > > > > > > > > Bets now being taken on what % will get done :) > > > > > > > > Thoughts? > > > > > > > > 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.
