Update gcl to the current release (bug fix) to build Maxima. Summary: 2.4.3 released Due to additional integer handling bugs that have come to light recently, 2.4.3 is being released. Maxima compilation as well as all known integer/bignum examples which failed to compile previously have been tested and are in working order. Summary: 2.4.2 released This is a bug fix only release meant as a stabilization of Dr. Schelter's earlier work. This will hopefully be a solid base for building maxima until the new 2.5.0 is released.
Created attachment 3361 [details] dev-lisp/gcl/gcl-2.4.3.ebuild
I'm not going to be able to squeeze this in before 1.4 is released.
I just tried the attached ebuild. You need a dependency: texi2html. (You probably already had that installed and never noticed it was used. I did not have it installed).
Hi guys Karl: I will, since this update seems to fix #7403 (assigned to me). Ole, Lars: I have committed and masked the ebuild, please test. Thanks for submission! George
I compiled Maxima and test it -- that was OK but there are some bugs in the installation of gcl-2.4.3: 1) a symlink: >>> /usr/bin/gcl.exe -> /var/tmp/portage/gcl-2.4.3/image//usr/lib/gcl-2.5.0/unixport/saved_gcl ll /usr/bin/gcl.exe lrwxrwxrwx 1 root root 70 Sep 4 22:04 /usr/bin/gcl.exe -> /var/tmp/portage/gcl-2.4.3/image//usr/lib/gcl-2.5.0/unixport/saved_gcl 2) loading of tkl.o Error: The function TK::GET-AUTOLOADS is undefined. --- and it depend on TCL/TK -- I have it installed: gcc -pipe -fwritable-strings -DVOL=volatile -I/var/tmp/portage/gcl-2.4.3/work/gcl-2.4.3/o -fsigned-char -c -march=i686 -O3 -pipe -I../o -I../h -I/usr/lib/../include -I/usr/lib/../include -I/usr/X11R6/include -O4 tkAppInit.c gcc -pipe -fwritable-strings -DVOL=volatile -I/var/tmp/portage/gcl-2.4.3/work/gcl-2.4.3/o -fsigned-char -c -march=i686 -O3 -pipe -I../o -I../h -I/usr/lib/../include -I/usr/lib/../include -I/usr/X11R6/include -O4 tkMain.c gcc -pipe -fwritable-strings -DVOL=volatile -I/var/tmp/portage/gcl-2.4.3/work/gcl-2.4.3/o -fsigned-char guis.o tkAppInit.o tkMain.o -o gcltkaux -L/usr/lib -ltk8.3 -L/var/tmp/portage/tk-8.3.3-r1/work/tk8.3.3/unix -ltk8.3 -L/usr/X11R6/lib -lX11 -I/usr/X11R6/include -L/usr/lib -ltcl8.3 `echo -ldl -lm | sed -e s:-lieee::g` -lm -lreadline -lncurses /usr/lib/libbfd.a /usr/lib/libiberty.a ../o/gcllib.a echo '(load "../tkl.o")(TK::GET-AUTOLOADS (directory "*.lisp"))' | ../../unixport/saved_gcl) GCL (GNU Common Lisp) Version(2.5.0) Wed Sep 4 22:04:15 CEST 2002 Licensed under GNU Library General Public License Contains Enhancements by W. Schelter > Loading ../tkl.o Error: free(3) error. Fast links are on: do (si::use-fast-links nil) for debugging Error signalled by EVAL. Broken at LOAD. Type :H for Help. >> Error: The function TK::GET-AUTOLOADS is undefined. Fast links are on: do (si::use-fast-links nil) for debugging Error signalled by EVALHOOK. Backtrace: system:universal-error-handler > EVALHOOK Broken at LOAD. >>make[1]: Leaving directory `/var/tmp/portage/gcl-2.4.3/work/gcl-2.4.3/gcl-tk' (cd info ; make)
I did more test with gcl-tk (gcc-2.95.4): The problem is with the build in '/var/tmp/portage/.../' This bug (feature) is also in /usr/lib/tkConfig.sh: TK_BUILD_LIB_SPEC='-L/var/tmp/portage/tk-8.3.3-r1/work/tk8.3.3/unix -ltk8.3' root]#gcl GCL (GNU Common Lisp) Version(2.5.0) Wed Sep 4 23:16:12 CEST 2002 Licensed under GNU Library General Public License Contains Enhancements by W. Schelter >(si::tkconnect) Loading /usr/lib/gcl-2.5.0/gcl-tk/tkl.o start address -T 839e000 Finished loading /usr/lib/gcl-2.5.0/gcl-tk/tkl.o "/usr/lib/gcl-2.5.0/gcl-tk/gcltksrv 127.0.0.1 5016 24472 localhost:0.0 "/usr/lib/gcl-2.5.0/gcl-tk/gcltksrv: /var/tmp/portage/gcl-2.4.3/image//usr/lib/gcl-2.5.0/gcl-tk/gcltkaux: No such file or directory /usr/lib/gcl-2.5.0/gcl-tk/gcltksrv: exec: /var/tmp/portage/gcl-2.4.3/image//usr/lib/gcl-2.5.0/gcl-tk/gcltkaux: cannot execute: No such file or directory the bug is in /usr/lib/gcl-2.5.0/gcl-tk/gcltksrv: GCL_TK_DIR=/var/tmp/portage/gcl-2.4.3/image//usr/lib/gcl-2.5.0/gcl-tk change it to: GCL_TK_DIR=/usr/lib/gcl-2.5.0/gcl-tk worked. I have update the ebuild by adding # fix the GCL_TK_DIR=/var/tmp/portage/gcl-2.4.3/image// mv ${D}/usr/lib/gcl-2.5.0/gcl-tk/gcltksrv ${D}/usr/lib/gcl-2.5.0/gcl-tk/gcltksrv.orig sed -e "s:${D}::g" < ${D}/usr/lib/gcl-2.5.0/gcl-tk/gcltksrv.orig > ${D}/usr/lib/gcl-2.5.0/gcl-tk/gcltksrv rm ${D}/usr/lib/gcl-2.5.0/gcl-tk/gcltksrv.orig chmod 0755 ${D}/usr/lib/gcl-2.5.0/gcl-tk/gcltksrv to the end of src_install(): Still TCL/TK (TK::GET-AUTOLOADS) problem and the demo is not working in this release of gcl -- see the readme This version is better that 2.4.0 and I can live with the TCL/TK problem.
Created attachment 3660 [details] update dev-lisp/gcl/gcl-2.4.3.ebuild fix the GCL_TK_DIR=/var/tmp/portage/gcl-2.4.3/image problem
Hi Lars Thanks for tracking some more issues. I updated the ebuild (still masked). BTW, I changed that absolute symlink by gcl.exe to relative (do we need this under Linux at all? From the name it seems to be some "compatibility" thing, though 'file' says that's an elf binary). I also added dependency on gmp, since I see that gcl supports it. George
It looks OK. For you information: I used/had installed a patched version of gmp-4.1-r1 (-r2) with bugfixes from the gmp homepage: 22,37d20 < < # mpz_class::get_num and get_den return a special type which behaves < # like mpz_class& but is not identical. < # An actual mpz_class& can be obtained with this patch. < # This is a change to the C++ ABI (inline code only) < patch < ${FILESDIR}/gmpxx.h.getnum.diff < < # mpz_powm_ui may return a result outside the range 0 to m-1 in < # certain circumstances (e==1, abs(b)>m, m normalized). < cd ${S}/mpz < patch < ${FILESDIR}/powm_ui.c.41.diff < < # mpf_inp_str returns an incorrect character count for valid input, < # and leaks memory for invalid input. < cd ${S}/mpf < patch < ${FILESDIR}/mpf_inp_str.c.diff Maxima (CVS) xmaxima + make check : Running test suite with gcl... *** Summary results for tests recorded in *** log file tests-gcl.log: Error summary: Error(s) found in rtest15.mac: (4) Expected failures (known bugs in this version of Maxima): rtest15.mac: (4) Timing: real time : 10.450 secs run time : 92.170 secs *** end of summary for tests-gcl.log
Hi Lars: thanks for testing! So, does this look now ready to go to everybody? I'll unmask it topmorrow if nobody will voice against till then. George
Unmasked, since there were no objections. Closing this bug now. Thanks everybody, especially Lars! George