Hi, Ok, will look deeper into that issue when time permits but right now I just checked out the SVN tree (rev. 2100) and the build process stops directly at an earlier stage :
===============================================================================
[undoing binding stack and other enclosing state... done]
[saving current Lisp image into
C:\MinGW\msys\1.0\home\gvanuxem\fricas_build\build\i686-pc-mingw32\bin\lisp.exe:
writing 3240 bytes from the read-only space at 22000000
writing 2560 bytes from the static space at 22100000
writing 30834688 bytes from the dynamic space at 22300000
done]
echo timestamp > do_it.sbcl
make[2]: Leaving directory `/home/gvanuxem/fricas_build/src/lisp'
make[2]: Entering directory `/home/gvanuxem/fricas_build/src/boot'
Building stage 0
[ -d stage0 ] || ../../../fricas/config/mkinstalldirs stage0
mkdir -p -- stage0
rm -rf prev-stage
rm -f stage0/ptyout.fasl stage0/btincl2.fasl stage0/btscan2.fasl
stage0/typrops.fasl stage0/btpile2.fasl stage0/typars.fasl
stage0/tytree1.fasl
rm -f stage0/ptyout.clisp stage0/btincl2.clisp stage0/btscan2.clisp
stage0/typrops.clisp stage0/btpile2.clisp stage0/typars.clisp
stage0/tytree1.clisp
make OBJECTS="stage0/ptyout.fasl stage0/btincl2.fasl stage0/btscan2.fasl
stage0/typrops.fasl stage0/btpile2.fasl stage0/typars.fasl
stage0/tytree1.fasl" stage0/bootsys.exe
make[3]: Entering directory `/home/gvanuxem/fricas_build/src/boot'
/home/gvanuxem/fricas_build/../fricas/src/scripts/build_helper
--translate_boot --debug=no --use=./prev-stage/bootsys
--output=../../../fricas/src/boot/compiled/ptyout.clisp
../../../fricas/src/boot/compiled/../../../fricas/src/boot/ptyout.boot
/home/gvanuxem/fricas_build/../fricas/src/scripts/build_helper: line 4:
./prev-stage/bootsys: No such file or directory
make[3]: *** [../../../fricas/src/boot/compiled/ptyout.clisp] Error 127
make[3]: Leaving directory `/home/gvanuxem/fricas_build/src/boot'
make[2]: *** [stage0/stamp_bootsys] Error 2
make[2]: Leaving directory `/home/gvanuxem/fricas_build/src/boot'
make[1]: *** [all-boot] Error 2
make[1]: Leaving directory `/home/gvanuxem/fricas_build/src'
make: *** [all-src] Error 2
==============================================================================
Log attached.
Reading the output, it *seems* the build machinery removes the prev-stage
directory ('rm -rf prev-stage') and after tries to use a file in it
'build_helper --translate_boot --debug=no --use=./prev-stage/bootsys'.
That's cryptic to me.
So I quickly tried to see which revision modified the build machinery
without success.
May be that speaks to someone...
I admit I should use Linux to test before but here I only have a USB key
with Linux on it. Not a big fan of the combination of cygwin.dll +
msvcrt.dll too.
Cheers
__
Greg
2016-09-01 13:57 GMT+02:00 Waldek Hebisch <[email protected]>:
> >
> > Nevertheless, the full source build seems unreliable on
> Windows/MinGW/MSYS.
> >
> > Yesterday I tried to build the full source out of the tree and the build
> > process stopped without giving me back the prompt, just an apparent
> pause.
> >
> > Don't ask me what was happened but I suspect a concurrency problem. I
> used
> > 'make -j4' and as far as I know :
> >
> > 1) SBCL threaded code is not supported, the message from SBCL is :
> WARNING:
> > the Windows port is fragile, particularly for multithreaded code.
> > Unfortunately, the development team currently lacks the time
> > and resources this platform demands.
>
> FriCAS does not use sbcl threads, so if there is any use of
> threads is due to internal sbcl code.
>
> > 2) the build process uses the port of make from MinGW/MSYS
> > 3) last and not least, from my point of view filesystem management on
> > Windows is particular and shows a lot of apparently concurrency strange
> > things (file not unlocked after closing an application which leads
> > sometimes to reboot or at least log out the session and this is just an
> > example).
> >
> > Now, and today I tried to build the full source with a simple 'make' and
> > out of the tree (from scratch) and it stopped with this :
> >
> > ===============================================================
> > AXIOM="C:/MinGW/msys/1.0/home/gvanuxem/fricas_build/target/
> i686-pc-mingw32"
> > spad-lib="C:/MinGW/msys/1.0/home/gvanuxem/fricas_build/
> target/i686-pc-mingw32/li
> > b/libspad.so"
> > foreign routines found
> > openServer result -2
> > FriCAS Computer Algebra System
> > Version: FriCAS 2016-08-28
> > Timestamp: Tue Aug 30 06:09:20 GMT 2016
> > ------------------------------------------------------------
> -----------------
> > Issue )copyright to view copyright notices.
> > Issue )summary for a summary of useful system commands.
> > Issue )quit to leave FriCAS and return to shell.
> > ------------------------------------------------------------
> -----------------
> >
> > Using local database ./rf/algebra/compress.daase.. Re-reading
> > compress.daas
> > e Using local database ./rf/algebra/interp.daase.. Re-reading
> > interp.daase
> > Using local database ./rf/algebra/operation.daase.. Re-reading
> > operation.da
> > ase
> > Using local database ./rf/algebra/category.daase.. Re-reading
> > category.daas
> > e
> > Using local database ./rf/algebra/browse.daase..
> > (1) -> Compiling FriCAS source code from file
> > C:/MinGW/msys/1.0/home/gvanuxem/fricas_build/src/algebra/INS.spad
> > using old system compiler.
> > INS abbreviates category IntegerNumberSystem
> > ------------------------------------------------------------------------
> > initializing NRLIB INS for IntegerNumberSystem
> > compiling into NRLIB INS
> >
> > >> System error:
> > The function BOOT::|IntegerNumberSystem| is undefined.
>
> AFAICS the direct cause of this error is due to limitation
> (you may call it a bug) in FriCAS: to compile category
> you need access to compiled form of other categories,
> sometimes even to the same category that you are trying
> to compile. However, at this stage all categories should
> have been compiled (in special mode which give somewhat incomplete
> code, but enough for the current stage) and consequently compiled
> form of INS should be available. So what may be wrong?
> The first guess is incorrect information in databases.
> Databases contain information about existing categories
> and if a category is missing from database, then it
> is not available for FriCAS. There could be some problem
> with paths, but then you probably would get different
> error message. First I would look if databases (in 'rf'
> subdirectory of build tree look sane (they should have
> the same content as during build on Linux)). The I
> would check if .fasl files are available in target
> tree ('C:/MinGW/msys/1.0/home/gvanuxem/fricas_build/target/
> i686-pc-mingw32/algebra').
>
> To debug such problem I would start interpsys by hand,
> starting it in the same directory and using the same
> values of environment variables as during build. Then
> I would try the offending ')compile' command by hand
> looking why it failed.
>
>
> --
> Waldek Hebisch
>
> --
> You received this message because you are subscribed to the Google Groups
> "FriCAS - computer algebra system" group.
> To unsubscribe from this group and stop receiving emails from it, send an
> email to [email protected].
> To post to this group, send email to [email protected].
> Visit this group at https://groups.google.com/group/fricas-devel.
> For more options, visit https://groups.google.com/d/optout.
>
--
You received this message because you are subscribed to the Google Groups
"FriCAS - computer algebra system" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to [email protected].
To post to this group, send email to [email protected].
Visit this group at https://groups.google.com/group/fricas-devel.
For more options, visit https://groups.google.com/d/optout.
log.gz
Description: GNU Zip compressed data
