On December 11, 2006 11:40 AM Gabriel Dos Reis wrote: > > On Mon, 11 Dec 2006, Bill Page wrote: > > | On December 10, 2006 11:41 PM Gabriel Dos Reis wrote: > | > ... > | > Axiom build-improvements branch revision 395 contains all the > | > necessary bits to build Axiom on windows and reproduce the > | > above. Please could you give it a try? > | > > | > | I pulled the latest patches to build-improvements and tried an > | out-of-source build in MSYS but > | > | mkdir axiom.test > | cd axiom.test > | ../build-improvements/configure > | > | hangs my Windows XP system during a confidence test. I am not > | able even to kill the process. > > That is very odd. >
Yes, it was VERY ODD indeed, but it turned out not to be a problem with MSYS/MinGW or build-improvements at all. See solution below. > | > | This same MSYS environment builds axiom--windows--1 and > | gcl-2.6.8pre with no problems. > | After a little more testing, I discovered that this problem was not limited to build-improvements. Contrary to what I said above the build of axiom--windows--1 and gcl also failed. > | Can you give me some more specifics about how you are testing > | this on windows? > > I pretty much do the same thing as you. > > (1) get the axiom source > (2) create a build directory > (3) invoke configure > (4) make > > Could you send me the (partial) config.log? > After spending most of the day on this problem, I finally determined that at random places during the configure (or in fact any long build operation involving MinGW) some process named 'LVPrcSrv.exe' was suddenly hogging 99% of all available CPU time. Google told me that this program was part of the Logitech QuikCam video effects package. In fact I had installed the new camera and software about a couple of months ago. And another search turned up reports of exactly this type of problem reported by cygwin users with a work-round of stopping the video effects service. (I have not yet been able to find a fix for this from Logitech.) The final result. As you asked, I was able to run: mkdir axiom.test cd axiom.test ../build-improvements/configure make up to this point: ... compiling SETCAT-.lsp to SETCAT-.o compiling SINT.lsp to SINT.o compiling STAGG.lsp to STAGG.o compiling STAGG-.lsp to STAGG-.o compiling SYMBOL.lsp to SYMBOL.o compiling TSETCAT.lsp to TSETCAT.o compiling TSETCAT-.lsp to TSETCAT-.o compiling UFD.lsp to UFD.o compiling UFD-.lsp to UFD-.o compiling ULSCAT.lsp to ULSCAT.o compiling UPOLYC.lsp to UPOLYC.o compiling UPOLYC-.lsp to UPOLYC-.o compiling URAGG.lsp to URAGG.o compiling URAGG-.lsp to URAGG-.o compiling VECTOR.lsp to VECTOR.o ===================================== === algebra bootstrap complete ====== ===================================== compiling AHYP.spad to AHYP.NRLIB /bin/sh: ./../..//build/i686-pc-mingw32/bin/interpsys: No such file or directory make[2]: *** [../..//int/algebra/AHYP.NRLIB/code.o] Error 127 make[2]: Leaving directory `/home/Administrator/axiom/src/algebra' make[1]: *** [all-algebra] Error 2 make[1]: Leaving directory `/home/Administrator/axiom/src' make: *** [all-src] Error 2 ------------- This looks very promising to me. I recall a problem similar to this in the axiom--windows--1 build. Did you catch the patch from axiom--windows--1 in boot where you need to ignore CR in windows line endings? I will look at this further tomorrow. Regards, Bill Page. _______________________________________________ Axiom-developer mailing list [email protected] http://lists.nongnu.org/mailman/listinfo/axiom-developer
