sbcl 1.0.29 fails to build on 10.6

[luis.beca@…]

sbcl 1.0.29 fails to build on Snow Leopard.

[toby@…]

needs a darwin 10 section

[fusion@…]

Cc Me!

[tonytung@…]

Cc Me!

[tm@…]

Cc Me!

[smanek@…]

Cc Me!

[pff@…]

Cc Me!

[nik@…]

fixes for snow leopard compatibility are in upstream sbcl version 1.0.31.2 ​http://git.boinkor.net/gitweb/sbcl.git?a=commit;h=9e0cc7f6afca5682944d9cc88eed63b5d80cbe9e

[m4k0t0y@…]

Cc Me!

[glen.henshaw@…]

Cc Me!

[macports@…]

Cc Me!

[hiroki.ora@…]

Cc Me!

[ronhuang@…]

Cc Me!

[easieste@…]

Cc Me!

[epimetheus314@…]

I made an ugly patch to Portfile. It tries to use source files from CVS if you have a 64bit CPU. I'm not sure it's a right choice, but anyway it seems to work for me. Also I attach a patch to make-config.sh which guesses (wrongly, in my case) the architecture.

[epimetheus314@…]

I forgot to add about the patches above -- I just wanted sbcl in a x86_64 binary in order to install maxima into my iMac (32bit SL kernel on 64bit Core2Duo). I didn't tested them on any other environment, so please test and revise.

[carlon.luca@…]

I still get this message when installing sbcl under MacOS Snow Leopard.

[allencmcbride@…]

Cc Me!

[allencmcbride@…]

the "snowleopard" keyword should be added, so that this will show up at SnowLeopardProblems. I guess the owner has to do that?

[snc@…]

Using the patches, iit asked me for a CVS password. I just hit return and it eventually failed in the build phase from a missing file.

Error: Target org.macports.build returned: shell command "unset LD_PREBIND && unset LD_PREBIND_ALLOW_OVERLAP && sh make.sh "/opt/local/var/macports/build/_Users_macports_sbcl/work/sbcl-1.0.29-x86_64-darwin/src/runtime/sbcl --core /opt/local/var/macports/build/_Users_macports_sbcl/work/sbcl-1.0.29-x86_64-darwin/output/sbcl.core --disable-debugger --sysinit /dev/null --userinit /dev/null" " returned error 127
Command output: sh: make.sh: No such file or directory

[brenton.kenkel@…]

Cc Me!

[arcadiorubiogarcia@…]

Cc Me!

[dpkatz@…]

Cc Me!

[earcar@…]

Cc Me!

[dominick.saputo@…]

Cc Me!

[epimetheus314@…]

I attached a revised patch to Portfile, using the latest cvs version. The patch for make-config.sh I attached before is no longer needed because it was integrated in the upstream. The patched Portfile should work for x86_64 SLs, but please note that:

(1) the installation process must be finished all at once, not be divided by {fetch, build, etc.} phases; i.e., just use 'port install sbcl'. This is because the version number for this cvs version is set only in the post-fetch phase.

(2) hit the return key if a cvs password is requested. I'm not sure about how to get rid of this.

(3) sometimes the installation process might stop by Segmentation fault, which I don't know why.

[macsforever2000@…]

Cc Me!

[achim.passen@…]

Cc Me!

[kuba@…]

Replying to luis.beca@…:

sbcl 1.0.29 fails to build on Snow Leopard.

Re CVS password: there's a typo in the patch. The CVS account is not set cvsroot ":pserver:anonimous@sbcl.cvs.sourceforge.net:/cvsroot/sbcl" It should be set cvsroot ":pserver:anonymous@sbcl.cvs.sourceforge.net:/cvsroot/sbcl" I'll post updated patch sometime.

[dersh@…]

Cc Me!

[macsforever2000@…]

Cc reporter of duplicate #22273.

[kuba@…]

Replying to epimetheus314@…:

(2) hit the return key if a cvs password is requested. I'm not sure about how to get rid of this.

The cvsroot should have a colon after anonymous, like so:

set cvsroot ":pserver:anonymous:@sbcl.cvs.sourceforge.net:/cvsroot/sbcl"

That way you provide an empty password.

[kuba@…]

I've uploaded a patch that updates sbcl to latest release 1.0.32, it works fine on SL. I guess the maintainer can roll it in and close this ticket.

[kuba@…]

.. that was a little bit too early, the build eventually barfs since two of the posix tests fail. I've patched those and are rebuilding now. I will attach a complete patch set soon.

[kuba@…]

sbcl 1.0.32 portfile, works on Snow Leopard, all tests pass

[kuba@…]

patch needed for posix tests in 1.0.32 to pass on SL

[kuba@…]

The most recent patchset I attaches compiles cleanly and just works ;) Methinks it's time to roll it in, this time for real (famous last words).

[kuba@…]

I just compiled maxima using today's sbcl portfile and on a cursory check it's OK: integrate(sin(x),x); gives -cos(x) as a result.

[luis.beca@…]

Replying to kuba@…:

.. that was a little bit too early, the build eventually barfs since two of the posix tests fail. I've patched those and are rebuilding now. I will attach a complete patch set soon.

I don't know if this is a useful observation but I was able to compile sbcl 1.0.32 with threads (and maxima) without problems, just using the portfile patch. However, just building sbcl does fail.

[kuba@…]

Replying to luis.beca@…:

Replying to kuba@…:

.. that was a little bit too early, the build eventually barfs since two of the posix tests fail. I've patched those and are rebuilding now. I will attach a complete patch set soon.

I don't know if this is a useful observation but I was able to compile sbcl 1.0.32 with threads (and maxima) without problems, just using the portfile patch. However, just building sbcl does fail.

The posix test fails, perhaps the way you compile it does not run the tests. The cvs version from epimetheus314's patch didn't have the test functionality fully implemented, thus it'd compile "without" problems.

[snc@…]

Cc Me!

[t.kunishi@…]

Cc Me!

[snc@…]

built for me …

[snc@…]

I've attached a Portfile change I just whipped up simply turning platform darwin 9 i386 into platform i386. Plugged in the right versions for things and it built.

Also, my archived build folder is available: ​http://tazamahal.com/macports/_opt_local_var_macports_packages_darwin_x86_64/sbcl-1.0.32_0+html+i386.x86_64.tbz2

[Martin.Buchmann@…]

Sorry for this kind of rookie question: If I try to installed sbcl using the regular way, i.e. just using port install sbcl without patching anything myself, I still end with version 1.0.30 and the initial error. Should I just wait a few more days or is there anything I missed so far.

[larsko@…]

Cc Me!

[jabial@…]

Cc Me!

[easieste@…]

sbcl-1.0.33 seems to work "out of the box" under OS X 10.6.2 x86_64.

Attaching the Portfile that works for me.

[easieste@…]

sbcl-1.0.33 portfile that works under x86_52-darwin-10.2.0

[arto.bendiken@…]

Cc Me!

[arto.bendiken@…]

The 1.0.33 portfile posted by easieste in comment #45 worked fine for me.

[bretthoerner@…]

Cc Me!

[bretthoerner@…]

Ditto, easieste's Portfile worked like a charm.

[krischik@…]

easieste uploaded his patch 3 weeks ago and the maintainer did nothing - I think we should declare the Port abandoned and let easieste commit his change.

[easieste@…]

Ok. Committing.

[easieste@…]

Final version committed as r61940.

Are we supposed to mark potentially abandoned ports in some manner?

[jabial@…]

When is it going to be available through selfupdate?

[jabial@…]

It has become avaiable through selfupdate now. Trying to compile.

[macsforever2000@…]

Replying to easieste@…:

Are we supposed to mark potentially abandoned ports in some manner?

You can file a ​port abandoned ticket.

[jabial@…]

Works ! Thank you. I think you can re-close it now.

[glen.henshaw@…]

Replying to jabial@…:

Sadly, the new port doesn't compile for me... error follows:

---> Computing dependencies for sbcl ---> Fetching sbcl ---> Attempting to fetch sbcl-1.0.33-source.tar.bz2 from ​http://internap.dl.sourceforge.net/sbcl ---> Attempting to fetch sbcl-1.0.12-x86-darwin-binary.tar.bz2 from ​http://internap.dl.sourceforge.net/sbcl ---> Verifying checksum(s) for sbcl ---> Extracting sbcl ---> Applying patches to sbcl ---> Configuring sbcl ---> Building sbcl Error: Target org.macports.build returned: shell command "unset LD_PREBIND && unset LD_PREBIND_ALLOW_OVERLAP && cd /opt/local/var/macports/build/_opt_local_var_macports_sources_rsync.macports.org_release_ports_lang_sbcl/work/sbcl-1.0.33 && sh make.sh "/opt/local/var/macports/build/_opt_local_var_macports_sources_rsync.macports.org_release_ports_lang_sbcl/work/sbcl-1.0.12-x86-darwin/src/runtime/sbcl --core /opt/local/var/macports/build/_opt_local_var_macports_sources_rsync.macports.org_release_ports_lang_sbcl/work/sbcl-1.0.12-x86-darwin/output/sbcl.core --disable-debugger --sysinit /dev/null --userinit /dev/null" " returned error 1 Command output: ; compiling (DEFUN SAMPLES-PERCENT ...) ; compiling (DEFUN PRINT-CALL-GRAPH-HEADER ...) ; compiling (DECLAIM (TYPE # ...)) ; compiling (DEFVAR *REPORT-SORT-BY* ...) ; compiling (DECLAIM (TYPE # ...)) ; compiling (DEFVAR *REPORT-SORT-ORDER* ...) ; compiling (DEFUN PRINT-FLAT ...) ; compiling (DEFUN PRINT-CYCLES ...) ; compiling (DEFUN PRINT-GRAPH ...) ; compiling (DEFUN REPORT ...) ; compiling (DEFUN SAMPLE-PC-FROM-PC-OR-OFFSET ...) ; compiling (DEFUN ADD-DISASSEMBLY-PROFILE-NOTE ...) ; compiling (PUSHNEW (QUOTE ADD-DISASSEMBLY-PROFILE-NOTE) ...) ; compiling (DEFUN PROFILE-CALL-COUNTS ...) ; compiling (DEFUN PROFILE-CALL-COUNTS-FOR-FUNCTION ...) ; compiling (DEFUN UNPROFILE-CALL-COUNTS ...) ; compiling (DEFUN ENABLE-CALL-COUNTING ...) ; compiling (DEFUN DISABLE-CALL-COUNTING ...) ; compiling (DEFUN ENABLE-CALL-COUNTING-FOR-FUNCTION ...) ; compiling (DEFUN TEST-0 ...) ; compiling (DEFUN TEST ...) ; compiling (PROVIDE (QUOTE SB-SPROF))

; SYS:CONTRIB;SB-SPROF;SB-SPROF.FASL.NEWEST written ; compilation finished in 0:00:01.067

• true

WARNING! Some of the contrib modules did not build successfully or pass their self-tests. Failed contribs:"

sb-posix

Error: Status 1 encountered during processing.

Any insight?

[easieste@…]

I've seen this happen before but it had gone away.

Somehow, sometimes the POSIX return values for various tests on the filesystem return different values then expected. I think the port itself has compiled ok, and would be okay to install.

Not sure how to track this down in a reproducible manner.

[arcadiorubiogarcia@…]

Cc Me!

[fusion@…]

I ran Disk Utility's fix permissions function and then tried rebuilding and got this error:

--->  Computing dependencies for sbcl
--->  Fetching sbcl
--->  Verifying checksum(s) for sbcl
--->  Extracting sbcl
--->  Applying patches to sbcl
--->  Configuring sbcl
--->  Building sbcl
Error: Target org.macports.build returned: shell command "unset LD_PREBIND && unset LD_PREBIND_ALLOW_OVERLAP && cd /opt/local/var/macports/build/_opt_local_var_macports_sources_rsync.macports.org_release_ports_lang_sbcl/work/sbcl-1.0.33 && sh make.sh "/opt/local/var/macports/build/_opt_local_var_macports_sources_rsync.macports.org_release_ports_lang_sbcl/work/sbcl-1.0.12-x86-darwin/src/runtime/sbcl --core /opt/local/var/macports/build/_opt_local_var_macports_sources_rsync.macports.org_release_ports_lang_sbcl/work/sbcl-1.0.12-x86-darwin/output/sbcl.core --disable-debugger --sysinit /dev/null --userinit /dev/null" " returned error 1
Command output: ; compiling (DEFUN SAMPLES-PERCENT ...)
; compiling (DEFUN PRINT-CALL-GRAPH-HEADER ...)
; compiling (DECLAIM (TYPE # ...))
; compiling (DEFVAR *REPORT-SORT-BY* ...)
; compiling (DECLAIM (TYPE # ...))
; compiling (DEFVAR *REPORT-SORT-ORDER* ...)
; compiling (DEFUN PRINT-FLAT ...)
; compiling (DEFUN PRINT-CYCLES ...)
; compiling (DEFUN PRINT-GRAPH ...)
; compiling (DEFUN REPORT ...)
; compiling (DEFUN SAMPLE-PC-FROM-PC-OR-OFFSET ...)
; compiling (DEFUN ADD-DISASSEMBLY-PROFILE-NOTE ...)
; compiling (PUSHNEW (QUOTE ADD-DISASSEMBLY-PROFILE-NOTE) ...)
; compiling (DEFUN PROFILE-CALL-COUNTS ...)
; compiling (DEFUN PROFILE-CALL-COUNTS-FOR-FUNCTION ...)
; compiling (DEFUN UNPROFILE-CALL-COUNTS ...)
; compiling (DEFUN ENABLE-CALL-COUNTING ...)
; compiling (DEFUN DISABLE-CALL-COUNTING ...)
; compiling (DEFUN ENABLE-CALL-COUNTING-FOR-FUNCTION ...)
; compiling (DEFUN TEST-0 ...)
; compiling (DEFUN TEST ...)
; compiling (PROVIDE (QUOTE SB-SPROF))

; SYS:CONTRIB;SB-SPROF;SB-SPROF.FASL.NEWEST written
; compilation finished in 0:00:01.637
* true

WARNING! Some of the contrib modules did not build successfully or pass
their self-tests. Failed contribs:"
  asdf-install

Error: Status 1 encountered during processing.

I also believe it would probably be OK to install at this point, though I'm a little worried that we are all having different failures in the test phase.

[glen.henshaw@…]

I also fixed permissions and tried to reinstall, and got the same error as before:

WARNING! Some of the contrib modules did not build successfully or pass their self-tests. Failed contribs:"

sb-posix

Error: Status 1 encountered during processing.

Also, I can't managed to force the port to install after the self-test failure - it always tries to rebuild, and bombs in the same spot, no matter what options or flags I use.

[easieste@…]

After investigation, I no longer expect that running Disk Utility will solve this problem for anyone (sorry for the false information). The problem really seems to lie in Mac OS X not returning the proper POSIX values, possibly due to inherited code from FreeBSD which recently patched this problem.

But it seems that sometimes these tests are working under Snow Leopard which is just plain odd.

Upstream SBCL has this bug logged as ​https://bugs.launchpad.net/sbcl/+bug/421779, so we might expect this to work in future SBCL releases.

I'm going to enable patch, bump the revision number, and push the port. Please report if this works.

[jabial@…]

Works for me. Anybody else ?

[easieste@…]

Some of the SB-POSIX tests *still* fail so we up the ante by macro'ing out the execution of the tests.

Added myself as maintainer here as well, which I am not too sure I have the time for, but I seem to have more than either waqar or gwright.

Oh, and sbcl-1.0.34 just rolled out the door.

[easieste@…]

With sbcl-1.0.33_04, I think SBCL is now working on Snow Leopard for everyone, or at least I have seen no further reports to the contrary. This was achieved by commenting out all SB-POSIX tests that failed until no more failing tests remained.

Upstream, it seems that sbcl-1.0.35 will contain the necessary fixes.

So, I am closing the ticket. If people are still experiencing installation failures for this port, please let me know.

[antoine@…]

SBCL 1.0.34 doesn't build under Snow Leopard when booted into 64 bit kernel (x86_64)

The build goes into an infinite loop on a call to munmap:

//entering make-host-1.sh
//building cross-compiler, and doing first genesis
This is SBCL 1.0.12, an implementation of ANSI Common Lisp.
More information about SBCL is available at <http://www.sbcl.org/>.

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.
munmap: Invalid argument
munmap: Invalid argument
munmap: Invalid argument
^C

Works fine when booted into the 32 bit kernel (i386)

[antoine@…]

SBCL 1.0.34 also doesn't build under Snow Leopard with variant: +threads

Build dies during application of: patch-base-target-features.diff

--->  Applying /opt/local/var/macports/sources/rsync.macports.org/release/ports/lang/sbcl/files/patch-base-target-features.diff
DEBUG: Environment: MACOSX_DEPLOYMENT_TARGET='10.6'
DEBUG: Assembled command: 'cd "/opt/local/var/macports/build/_opt_local_var_macports_sources_rsync.macports.org_release_ports_lang_sbcl/work/sbcl-1.0.34" && /usr/bin/patch -p0'
patching file base-target-features.lisp-expr
Hunk #1 FAILED at 157.
Hunk #2 succeeded at 229 (offset 62 lines).
Hunk #3 succeeded at 406 (offset 69 lines).
1 out of 3 hunks FAILED -- saving rejects to file base-target-features.lisp-expr.rej
Error: Target org.macports.patch returned: shell command " cd "/opt/local/var/macports/build/_opt_local_var_macports_sources_rsync.macports.org_release_ports_lang_sbcl/work/sbcl-1.0.34" && /usr/bin/patch -p0 < '/opt/local/var/macports/sources/rsync.macports.org/release/ports/lang/sbcl/files/patch-base-target-features.diff'" returned error 1
DEBUG: Backtrace: shell command " cd "/opt/local/var/macports/build/_opt_local_var_macports_sources_rsync.macports.org_release_ports_lang_sbcl/work/sbcl-1.0.34" && /usr/bin/patch -p0 < '/opt/local/var/macports/sources/rsync.macports.org/release/ports/lang/sbcl/files/patch-base-target-features.diff'" returned error 1
    while executing
"command_exec patch "" "< '$patch'""
    (procedure "portpatch::patch_main" line 33)
    invoked from within
"$procedure $targetname"
Warning: the following items did not execute (for sbcl): org.macports.activate org.macports.patch org.macports.configure org.macports.build org.macports.destroot org.macports.install
Error: Status 1 encountered during processing.
To report a bug, see <http://guide.macports.org/#project.tickets>

[antoine@…]

base-target-features.lisp-expr.rej :

***************
*** 157,163 ****
   ;; and the %fs segment register.  Note that no consistent effort to
   ;; audit the SBCL library code for thread safety has been performed,
   ;; so caveat executor.
-  ; :sb-thread
  
   ;; lutex support
   ;;
--- 157,163 ----
   ;; and the %fs segment register.  Note that no consistent effort to
   ;; audit the SBCL library code for thread safety has been performed,
   ;; so caveat executor.
+ :sb-thread
  
   ;; lutex support
   ;;

[easieste@…]

Attaching a diff that gets around the misapplied patch problem, but I (again) encounter problems with failing sb-posix test.

This issue should really be tracked in another ticket, as the sbcl-1.0.34 commit by gwright was after the work I did to get sbcl-1.0.33 working.

[easieste@…]

Fix for +threads variant for sbcl-1.0.34

[easieste@…]

But the failing tests seem to be better handled by removing the patch-posix-tests.diff in the Portfile at this point.

[team@…]

Cc Me!

[team@…]

every time I'm trying to install sbcl with macports, version 1.0.30 is used. how can I instruct it to fetch 1.035? this is what I'm doing:

$ sudo port -d install sbcl@1.0.35 +threads
...
--->  Fetching sbcl
DEBUG: Executing org.macports.fetch (sbcl)
--->  Verifying checksum(s) for sbcl
DEBUG: Executing org.macports.checksum (sbcl)
--->  Checksumming sbcl-1.0.30-source.tar.bz2
DEBUG: Correct (md5) checksum for sbcl-1.0.30-source.tar.bz2
…

please help.

[jmr@…]

Trac is not tech support. Ask in a more appropriate venue. http://www.macports.org/contact.php