UPDATE: arj-3.10.21 -> 3.10.22 with many bugfixes and stylistic improvements
| Reported by: |
boeyms@… |
Owned by: |
boeyms@… |
|
Priority:
|
Normal
|
Milestone:
|
|
|
Component:
|
ports
|
Version:
|
1.4.42
|
|
Keywords:
|
arj update
|
Cc:
|
boeyms@…, jc@…, pipping@…, majick@…
|
|
Port:
|
|
|
|
So many changes: I couldn't bring myself to go through the pain of undoing them
and splitting the commits, so I hope that the description below is clear
enough.
- Update arj to latest version (3.10.22) (closes #11629);
- Remove unnecessary bin:gmake:gmake in depends_build (closes #7070);
- Update bin:autoconf:autoconf in depends_build to port:autoconf;
- Replace custom pre-configure script with standard syntax (use_autoconf and
autoconf.dir);
- Replace custom configure script with standard syntax (configure.dir and
removal of --prefix switch that is added by default);
- Replace custom pre-build script with standard syntax (prepend "prepare" to
build.target list);
- Remove custom destroot script (source now generates a Makefile that handles
"make install" for us);
- Patch postproc.c so that it runs as intended on the ARJ executables
(integrity_pattern[] differs from that in register.c and arj_proc.c, with
comment in latter indicating that it contains the correct array);
- Patch configure.in so that arjcrypt builds as a dynamic library is it is
supposed to (previously it built as an executable):
- this obsoletes the patch to arjcrypt.c, which is therefore removed;
- this necessitates changing DLL_FLAGS in makefile.in to prevent it from
sending a flag to Mac OS X libtool that it does not support (-s for
stripping symbols);
- it does so in proper style (by detecting the OS), which requires the
post-extract step of replacing the very old config.sub and config.guess
scripts included in the source with the ones provided by XCode, and
obsoletes the old patch to configure.in that comments out auto-detection of
the host;
- Patch makefile.in to not run postproc on arjcrypt.dylib, as doing so makes
the build fail, and it seems neither necessary nor appropriate to do so;
- Remove patch to COPT definition in makefile.in as it is not necessary once
config.guess and config.sub are recent enough to allow configure to guess the
OS.
I will commit this myself if I get no feedback from the maintainer within a week.
Closed in r26877.