sbcl: Failed contribs: sb-introspect

[bpabbott@…]

On Leopard running on a G5, I see ...

$ sudo port install sbcl
--->  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_www.macports.org_files_ports_lang_sbcl/work/sbcl-1.0.30 && sh make.sh "/opt/local/var/macports/build/_opt_local_var_macports_sources_www.macports.org_files_ports_lang_sbcl/work/sbcl-1.0.22-powerpc-darwin/src/runtime/sbcl --core /opt/local/var/macports/build/_opt_local_var_macports_sources_www.macports.org_files_ports_lang_sbcl/work/sbcl-1.0.22-powerpc-darwin/output/sbcl.core --disable-debugger --sysinit /dev/null --userinit /dev/null" " returned error 1
Command output: ;       The result is a (VALUES (INTEGER 1 576460752303423487) &OPTIONAL), not a (VALUES
;                                                                                 FIXNUM
;                                                                                 &REST
;                                                                                 T).
;       etc.

; compiling (DEFUN TEST-0 ...)
; compiling (DEFUN TEST ...)
; compiling (PROVIDE (QUOTE SB-SPROF))
; file: /opt/local/var/macports/build/_opt_local_var_macports_sources_www.macports.org_files_ports_lang_sbcl/work/sbcl-1.0.30/contrib/sb-sprof/sb-sprof.lisp
; in: DEFUN START-PROFILING
;     #'SB-SPROF::THREAD-DISTRIBUTION-HANDLER
; 
; caught STYLE-WARNING:
;   undefined function: THREAD-DISTRIBUTION-HANDLER
; 
; compilation unit finished
;   Undefined function:
;     THREAD-DISTRIBUTION-HANDLER
;   caught 1 STYLE-WARNING condition
;   printed 1 note


; SYS:CONTRIB;SB-SPROF;SB-SPROF.FASL.NEWEST written
; compilation finished in 0:00:02.529
* true

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

Error: Status 1 encountered during processing.

[jmroot (Joshua Root)]

Please remember to use Preview and WikiFormatting in future.

[yaseppochi (Stephen J. Turnbull)]

I get the same log on PPC Tiger (10.4.11). I do not have a problem on Intel Leopard (10.5.8). I guess it's a PowerPC issue, then.

[yaseppochi (Stephen J. Turnbull)]

Replying to stephen@…:

I get the same log on PPC Tiger (10.4.11). I do not have a problem on Intel Leopard (10.5.8). I guess it's a PowerPC issue, then.

FWIW it is due to a failure of a new test in sbcl-1.0.30/contrib/sb-introspect/test-driver.lisp, the one that looks like this:

(tai #'cons :heap
     #+(and (not ppc) gencgc)
     ;; FIXME: This is the canonical GENCGC result, the one below for PPC is
     ;; what we get there, but :LARGE T doesn't seem right. Figure out what's
     ;; going on.
     '(:space :dynamic :generation 6 :write-protected t :pinned nil :large nil)
     #+(and ppc gencgc)
     '(:space :dynamic :generation 6 :write-protected t :pinned nil :large t)
     ;; FIXME: Figure out what's the right cheney-result, and which platforms
     ;; return something else. The SPARC version here is what we get there,
     ;; but quite possibly that is the result on all non-GENCGC platforms.
     #+(and sparc (not gencgc))
     '(:space :read-only)
     #+(and (not sparc) (not gencgc))
     '(:space :dynamic))

I have no clue what's going on, but commenting out that form gives me a build on Tiger/PowerPC. Since that test is not present in sbcl 1.0.29, no other tests fail, and the sbcl developers themselves express a bit of mystification, I suppose patching out that test and using 1.0.30 on PPC is as safe as using 1.0.29. I have *not* attempted a build with that test in 1.0.29, though. I suppose it could pass there....

[jmroot (Joshua Root)]

Duplicate #23388 reports similar failure on x86.

[easye]

I don't have access to Leopard/G5, so helping out with this is a bit problematic.

The lang/sbcl port is now at 1.0.33. Could the people reporting failure try the new version of the port to see if they stil see failure?

[bpabbott@…]

Replying to easieste@…:

I don't have access to Leopard/G5, so helping out with this is a bit problematic.

The lang/sbcl port is now at 1.0.33. Could the people reporting failure try the new version of the port to see if they stil see failure?

Unfortunately, it didn't work for me.

$ sudo port install -d sbcl
Password:
--->  Computing dependencies for 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_www.macports.org_files_ports_lang_sbcl/work/sbcl-1.0.33 && sh make.sh "/opt/local/var/macports/build/_opt_local_var_macports_sources_www.macports.org_files_ports_lang_sbcl/work/sbcl-1.0.22-powerpc-darwin/src/runtime/sbcl --core /opt/local/var/macports/build/_opt_local_var_macports_sources_www.macports.org_files_ports_lang_sbcl/work/sbcl-1.0.22-powerpc-darwin/output/sbcl.core --disable-debugger --sysinit /dev/null --userinit /dev/null" " returned error 1
Command output: ;       The result is a (VALUES (INTEGER 1 576460752303423487) &OPTIONAL), not a (VALUES
;                                                                                 FIXNUM
;                                                                                 &REST
;                                                                                 T).
;       etc.

; compiling (DEFUN TEST-0 ...)
; compiling (DEFUN TEST ...)
; compiling (PROVIDE (QUOTE SB-SPROF))
; file: /opt/local/var/macports/build/_opt_local_var_macports_sources_www.macports.org_files_ports_lang_sbcl/work/sbcl-1.0.33/contrib/sb-sprof/sb-sprof.lisp
; in: DEFUN START-PROFILING
;     #'SB-SPROF::THREAD-DISTRIBUTION-HANDLER
;
; caught STYLE-WARNING:
;   undefined function: THREAD-DISTRIBUTION-HANDLER
;
; compilation unit finished
;   Undefined function:
;     THREAD-DISTRIBUTION-HANDLER
;   caught 1 STYLE-WARNING condition
;   printed 1 note


; SYS:CONTRIB;SB-SPROF;SB-SPROF.FASL.NEWEST written
; compilation finished in 0:00:02.983
* true

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

Error: Status 1 encountered during processing.
Before reporting a bug, first run the command again with the -d flag to get complete output.

[yaseppochi (Stephen J. Turnbull)]

I still see this failure with 1.0.36 on Mac OS X 10.4.11 (PPC)

The attached patch-introspect-tests.diff wins for me, but I don't know why. If somebody could report upstream, I would appreciate it, I don't have time or energy for the next couple of weeks to deal with yet another bug tracking system.

[easye]

Does sbcl-1.0.43 still fail?

[jmroot (Joshua Root)]

Current version builds on 10.5/PPC.