Opened 8 years ago

Closed 3 years ago

Last modified 3 years ago

#35543 closed defect (wontfix)

SuiteSparse fails to build on Tiger, G3

Reported by: tobia.tesan@… Owned by: michaelld (Michael Dickens)
Priority: Normal Milestone:
Component: ports Version: 2.1.2
Keywords: tiger Cc: zan@…
Port: SuiteSparse

Description (last modified by ryandesign (Ryan Schmidt))

:info:destroot /usr/libexec/gcc/powerpc-apple-darwin8/4.0.1/libtool: both -all_load and -noall_load can't be specified

Something with flags in a script, apparently.

Attachments (2)

main.log (24.8 KB) - added by tobia.tesan@… 8 years ago.
Build log
Portfile.diff (2.2 KB) - added by zan@… 8 years ago.

Download all attachments as: .zip

Change History (10)

Changed 8 years ago by tobia.tesan@…

Attachment: main.log added

Build log

comment:1 Changed 8 years ago by ryandesign (Ryan Schmidt)

Description: modified (diff)
Keywords: g3 SuiteSparse removed

comment:2 Changed 8 years ago by jmroot (Joshua Root)

Cc: ram@… zan@… added

Why is -Wl,-all_load needed in the first place?

comment:3 Changed 8 years ago by zan@…

-all_load tells ld to pull all files from the static archives. Without it you get an empty dylib. The alternative is to use 'ar -x' to extract all *.o from *.a into a temp dir, build .dylib from *.o, then clean up the temp dir. I'll test it out.

comment:4 Changed 8 years ago by zan@…

I've updated the Portfile, diff file attached. Several updates:

  1. Newer version 4.0.0 -> 4.0.2
  2. use ar -x to extract *.o and then compile dylib from *.o
  3. metis is a depend_lib, not a depend_build, because it is linked dynamically.

There are still 2 old patch files in the files dir ( and makefiles.patch) that can be removed. They are not being used.

Last edited 8 years ago by zan@… (previous) (diff)

Changed 8 years ago by zan@…

Attachment: Portfile.diff added

comment:5 Changed 8 years ago by skymoo (Adam Mercer)

Cc: ram@… removed

comment:6 Changed 4 years ago by mf2k (Frank Schima)

Owner: changed from macports-tickets@… to michaelld
Status: newassigned

Assigning to current maintainer.

comment:7 Changed 3 years ago by kencu (Ken)

Resolution: wontfix
Status: assignedclosed

I presume we can rust-bucket this as a wontfix by this point :)

comment:8 Changed 3 years ago by michaelld (Michael Dickens)

seems likely! maybe try to revisit if/once #47903 is dealt with?

Note: See TracTickets for help on using tickets.