Re: [fricas-devel] Re: build failure using clisp
Bill Page
Do you recall why you explicity set the following option:
./makemake --without-dynamic-ffi ...
in clisp-2.41.p11/spkg-install ? Is it required for some other parts of Sage?
For the purposes of building the full version of Axiom using clisp it
is necessary to allow '--with-dynamic-ffi'.
Regards,
Bill Page.
On 12/5/07, Waldek Hebisch <heb...@math.uni.wroc.pl> wrote:
>
> Bill Page wrote:
> > While trying to build fricas:
> >
> > page@sage:~$ cd ~/fricas-src
> <snip>
> > page@sage:~/fricas-build$ cd ~/fricas-build
> >
> > using clisp:
> >
> <snip>
> > Features:
> > (REGEXP SYSCALLS I18N LOOP COMPILER CLOS MOP CLISP ANSI-CL COMMON-LISP LISP=CL
> > INTERPRETER SOCKETS GENERIC-STREAMS LOGICAL-PATHNAMES SCREEN GETTEXT UNICODE
> > BASE-CHAR=CHARACTER PC386 UNIX)
>
> My version of Clisp gives:
> Features:
> (READLINE REGEXP SYSCALLS I18N LOOP COMPILER CLOS MOP CLISP ANSI-CL COMMON-LISP
> LISP=CL INTERPRETER SOCKETS GENERIC-STREAMS LOGICAL-PATHNAMES SCREEN FFI
> GETTEXT UNICODE BASE-CHAR=CHARACTER PC386 UNIX)
>
> Note that FFI is absent in version on sage, but present in my (standard)
> version.
>
> > C Modules: (clisp i18n syscalls regexp)
> > Installation directory: /home/page/sage-2.8/local/lib/clisp/
> > User language: ENGLISH
> > Machine: X86_64 (X86_64) sage.math.washington.edu [128.208.160.191]
> >
> > page@sage:~/fricas-build$ ../fricas-src/configure
> > --prefix=/home/page/local --with-lisp=clisp
> > ...
> > page@sage:~/fricas-build$ make
> >
> > I get the following error:
> >
> > ...
> > Bootstrap object copy
> > Checking for foreign routines
> > AXIOM="/home/page/fricas-build/target/x86_64-unknown-linux"
> > spad-lib="/home/page/fricas-build/target/x86_64-unknown-linux/lib/libspad.so"
> > sock-fasl="/home/page/fricas-build/target/x86_64-unknown-linux/autoload/ffi-func2.lisp"
> > foreign routines found
> > ;; Loading file
> > /home/page/fricas-build/target/x86_64-unknown-linux/autoload/ffi-func2.lisp
> > ...
> > *** - READ from
> > #<INPUT BUFFERED FILE-STREAM CHARACTER
> > #P"/home/page/fricas-build/target/x86_64-unknown-linux/autoload/ffi-func2.lisp"
> > @4>
> > : there is no package with name "FFI"
>
> Yes, apparently FFI is not present. Graphic and HyperDoc require
> FFI. I am not sure if we should seriously support clisp without FFI...
> We could try to detect lack of FFI in installed clisp (but we do not
> try to detect lack of FFI in other Lisps, where in theory one can also
> omit FFI).
>
> --
> Waldek Hebisch
> heb...@math.uni.wroc.pl
>
> >
>
William Stein
> William,
>
> Do you recall why you explicity set the following option:
>
> ./makemake --without-dynamic-ffi ...
>
> in clisp-2.41.p11/spkg-install ? Is it required for some other parts of Sage?
>
> For the purposes of building the full version of Axiom using clisp it
> is necessary to allow '--with-dynamic-ffi'.
Yes, I did that because when I was putting clisp into Sage (over two years ago),
that option was needed in order to even get clisp to build on many machines.
I don't remember why there was so much trouble building clisp without
that option.
If you can remove that option *and* get the clisp spkg to build on the
sage-supported
platforms (as defined in the readme), then the dynamic ffi could be added back
into the Sage lisp.
-- william
mabshoff
On Dec 5, 6:06 pm, "William Stein" <wst...@gmail.com> wrote:
>
> > William,
>
> > Do you recall why you explicity set the following option:
>
> > ./makemake --without-dynamic-ffi ...
>
> > in clisp-2.41.p11/spkg-install ? Is it required for some other parts of Sage?
>
> > For the purposes of building the full version of Axiom using clisp it
> > is necessary to allow '--with-dynamic-ffi'.
>
> Yes, I did that because when I was putting clisp into Sage (over two years ago),
> that option was needed in order to even get clisp to build on many machines.
> I don't remember why there was so much trouble building clisp without
> that option.
not run makemake with the recommended options. I "fixed" that once a
couple months ago, but we reverted the changes because of the above
mentioned breakage on OSX.
clisp 2.43 is out and updating clisp has been on my lisp of things to
do for a while. (see #1002)
> If you can remove that option *and* get the clisp spkg to build on the
> sage-supported
> platforms (as defined in the readme), then the dynamic ffi could be added back
> into the Sage lisp.
>
> -- william
Cheers,
Michael
Bill Page
>
> The build fails on OSX without that option. Interestingly enough we do
> not run makemake with the recommended options. I "fixed" that once a
> couple months ago, but we reverted the changes because of the above
> mentioned breakage on OSX.
>
> clisp 2.43 is out and updating clisp has been on my lisp of things to
> do for a while. (see #1002)
>
> > If you can remove that option *and* get the clisp spkg to build on the
> > sage-supported platforms (as defined in the readme), then the dynamic
> > ffi could be added back into the Sage lisp.
> >
>
I tried that but the modified 'clisp-2.41.p11.spkg' does not build
apparently because the src directory does not contain everything
necessary to build ffi.
> We should do that during the update to 2.43. Any volunteers?
>
Well, so far I can confirm only that the following modified spkg:
http://sage.math.washington.edu/home/page/clisp-2.43-alpha.spkg
which is based on clisp-2.43 as distributed by the clisp project,
installs in Sage running on sage.math and it can re-build (-f ...)
maxima-5.13.0.p1, axiom4sage-0.3.1 and the full version of FriCAS
(rev: 134).
I did only a fairly mimimal change to 'clisp-2.43-alpha.spkg' to
eliminate all patches and adapt to the slightly changed build process
(no intermediate makemake step necessary). FFI is automatically
included. It it quite possible that this version may not build on OSX
etc. however I know that the clisp developers have made some
significant improvements in the build since 2.41. I would be glad in
anyone can try this 'clisp-2.43-alpha.spkg' and let me know what works
and what doesn't. Also please feel free to take the above and run with
it. ;-)
Regards,
Bill Page.
mabshoff
On Dec 5, 8:58 pm, "Bill Page" <bill.p...@newsynthesis.org> wrote:
> On 12/5/07, mabshoff wrote:
>
> > The build fails on OSX without that option. Interestingly enough we do
> > not run makemake with the recommended options. I "fixed" that once a
> > couple months ago, but we reverted the changes because of the above
> > mentioned breakage on OSX.
>
> > clisp 2.43 is out and updating clisp has been on my lisp of things to
> > do for a while. (see #1002)
>
> > On Dec 5, 6:06 pm, William Stein wrote:
> > > If you can remove that option *and* get the clisp spkg to build on the
> > > sage-supported platforms (as defined in the readme), then the dynamic
> > > ffi could be added back into the Sage lisp.
>
> I tried that but the modified 'clisp-2.41.p11.spkg' does not build
> apparently because the src directory does not contain everything
> necessary to build ffi.
>
> > We should do that during the update to 2.43. Any volunteers?
>
> Well, so far I can confirm only that the following modified spkg:
>
> http://sage.math.washington.edu/home/page/clisp-2.43-alpha.spkg
against it and will run doctests right after.
It seems to be about 3 mb larger than the previous spkg. Can we shrink
it a little more? I don't know what was removed last time.
>
> which is based on clisp-2.43 as distributed by the clisp project,
> installs in Sage running on sage.math and it can re-build (-f ...)
> maxima-5.13.0.p1, axiom4sage-0.3.1 and the full version of FriCAS
> (rev: 134).
>
> I did only a fairly mimimal change to 'clisp-2.43-alpha.spkg' to
> eliminate all patches and adapt to the slightly changed build process
> (no intermediate makemake step necessary). FFI is automatically
> included. It it quite possible that this version may not build on OSX
> etc. however I know that the clisp developers have made some
> significant improvements in the build since 2.41. I would be glad in
> anyone can try this 'clisp-2.43-alpha.spkg' and let me know what works
> and what doesn't. Also please feel free to take the above and run with
> it. ;-)
10.4 to see if everything seems to be working there, too.
>
> Regards,
> Bill Page.
Thanks a lot,
Michael