Steel Bank Common Lisp

On Mon, 2005-08-29 at 13:25 -0400, David Steuber wrote:
> On Aug 28, 2005, at 9:00 AM, R. Mattes wrote:
>=20
> > BTW, Do not debug a build process by manually CDing into a subdirectory
> > and invoking make. These makefiles depend on environment variables set
> > from the main makefile.
> >
> > HTH Ralf Mattes
>=20
> I've attempted to build the docs for 0.9.4.0 without success:
>=20
> cd doc/manual && make

As i allready said, you can't just 'cd' into a subdirectory and
expect your build to work ...

> ...
>=20
> (/home/david/usr/src/sbcl/doc/manual/docstrings/macro-sb-ext-with-=20
> unlocked-pack
> ages.texinfo) [69]) (./threading.texinfo Chapter 10 [70]
> ./threading.texinfo:32: I can't find file =20
> `struct-sb-thread-thread.texinfo'.
> @temp ->@input struct-sb-thread-thread.texinfo
>=20
> @includezzz ...mp {@input #1 }@expandafter }@temp
>                                                    @popthisfilestack
> l.32 @include struct-sb-thread-thread.texinfo
>=20
> Please type another input file name:
> [I hit enter each time this comes up]
>=20
> ...
>=20
> /home/david/usr/src/sbcl/doc/manual//threading.texinfo:173: @include =20
> `struct-sb-thread-waitqueue.texinfo': No such file or directory.
> /home/david/usr/src/sbcl/doc/manual//threading.texinfo:173: @include =20
> `fun-sb-thread-make-waitqueue.texinfo': No such file or directory.
> /home/david/usr/src/sbcl/doc/manual//threading.texinfo:173: @include =20
> `fun-sb-thread-waitqueue-name.texinfo': No such file or directory.
> /home/david/usr/src/sbcl/doc/manual//threading.texinfo:175: @include =20
> `fun-sb-thread-condition-broadcast.texinfo': No such file or directory.
> ../../contrib//sb-bsd-sockets/sb-bsd-sockets.texinfo:72: @include =20
> `fun-sb-bsd-sockets-socket-open-p.texinfo': No such file or directory.
> makeinfo: Removing output file =20
> `/home/david/usr/src/sbcl/doc/manual/sbcl.info' due to errors; use =20
> --force to preserve.
> make: *** [sbcl.info] Error 1
> rm sbcl.dvi asdf.dvi
> david@...
> $

As far as i understand these files are generated by the build process.
If you run from the doc/manual directory your Makefile lacks a lot of
important environment variables. Have a look at make-tempfiles.sh which
is supposed to create these texinfo files from the docstrings of the
running image.=20


> What am I doing wrong?  I was successful building at least the html =20
> docs over a year ago now.  I would like to have an up to date copy.

Not using make the way it's supposed to be uses ? ;-}

 Cheers RalfD

> System info: Debian/Sarge
> $ uname -a
> Linux apostrophe 2.6.11.9-686 #1 Thu Aug 25 04:40:51 EDT 2005 i686 =20
> GNU/Linux
> $ cat customize-target-features.lisp
> (lambda (list)
>    (pushnew :sb-thread list))
> $ sbcl
> This is SBCL 0.9.4, an implementation of ANSI Common Lisp.
> More information about SBCL is available at <http://www.sbcl.org/>.
>=20
> SBCL is free software, provided as is, with absolutely no warranty.
> It is mostly in the public domain; some portions are provided under
> BSD-style licenses.  See the CREDITS and COPYING files in the
> distribution for more information.
> * *features*
>=20
> (:SB-THREAD :ANSI-CL
>              :COMMON-LISP
>              :SBCL
>              :UNIX
>              :SB-DOC
>              :SB-TEST
>              :SB-PACKAGE-LOCKS
>              :SB-UNICODE
>              :IEEE-FLOATING-POINT
>              :X86
>              :ELF
>              :LINUX
>              :GENCGC
>              :STACK-GROWS-DOWNWARD-NOT-UPWARD
>              :C-STACK-IS-CONTROL-STACK
>              :STACK-ALLOCATABLE-CLOSURES
>              :LINKAGE-TABLE
>              :OS-PROVIDES-DLOPEN
>              :OS-PROVIDES-DLADDR)
> *
>=20

On 28. Aug 2005, at 22:17, R. Mattes wrote:

> On Mon, 2005-08-29 at 13:25 -0400, David Steuber wrote:
>
>> On Aug 28, 2005, at 9:00 AM, R. Mattes wrote:
>>
>>
>>> BTW, Do not debug a build process by manually CDing into a  
>>> subdirectory
>>> and invoking make. These makefiles depend on environment  
>>> variables set
>>> from the main makefile.
>>>
>>> HTH Ralf Mattes
>>>
>>
>> I've attempted to build the docs for 0.9.4.0 without success:
>>
>> cd doc/manual && make
>>
>
> As i allready said, you can't just 'cd' into a subdirectory and
> expect your build to work ...

Sound advice in general, but the documentation is meant to be built  
that way, and the message at the end of a successful build says so.

The doc build process starts a freshly-built in-tree sbcl that then  
writes out documentation strings, adding some texinfo markup.  I ran  
a quick check and wanted to say "worksforme", but after "make clean"  
the process breaks for me as well.  I'll investigate tomorrow.

Rudi

Joshua Boyd <jdboyd@...> writes:

> I previously built 0.9.3 on the same platform using 0.8.18 and gcc 3.3,
> now I am trying to build 0.9.4 using 0.9.3 and gcc 4.0.1.  Where the
> build fails is on the file sparc-arch.c, lines 89 and 90, with the
> errors illegal lvalue.  The lines in question read:
>     ((char *) *os_context_pc_addr(context)) = ((char *)*os_context_npc_addr(context));
>     ((char *) *os_context_npc_addr(context)) += 4;
>
> Since the sbcl file in question is unchanged between the two versions,
> I'm wondering if this is a known GCC4 issue?

I believe this is a known sbcl-with-gcc4 issue (gcc4 is correct in
complaining about those lines).  A patch which fixes the issue for you
would be welcome.

> Also, what is desired for contributing a binary package to the SBCL
> project for distribution.  The sourceforge site seems to only have
> 0.8.18 for download still, and I'd like to contribute a newer build
> (either 0.9.3 or this one when I get the issues worked out).  What is
> the proper testing for builds before distributing them, and would there
> be interest in solaris packages rather than just tarballs?

I can't comment on packages vs. tarballs, but I'm happy for now to
build sbcl binaries from time to time (though not necessarily every
month -- I don't believe that that's necessary, given that building
sbcl, even for users, should be straightforward.)

> Does lack of updates here mean that no one else is using this port anymore?

Coincidentally (and it has to be coincidence, given how long it takes
for my machine to build SBCL :-) I've just uploaded a binary of SBCL
for SPARC/SunOS.  Not that I'm using it, but it still builds for me
when I don't change anything on the machine in question.

Cheers,

Christophe