#42925 closed submission (fixed)
libgrss @0.5.0 new Portfile
Reported by: | juanrgar (Juan R. García Blanco) | Owned by: | dbevans (David B. Evans) |
---|---|---|---|
Priority: | Normal | Milestone: | |
Component: | ports | Version: | |
Keywords: | Cc: | bob4job@… | |
Port: | libgrss |
Description
tracker @0.17.6 depends on libgrss @0.5.0
I'm having issues with this port, although it all seems correct. First of all, during configure phase it does not recognise "--disable-silent-rules". Also, during post-activate phase, scrollkeeper shows also an error: it is like the command name is passed as argument to that same command.
Attachments (3)
Change History (14)
Changed 11 years ago by juanrgar (Juan R. García Blanco)
Attachment: | libgrss-Portfile.tgz added |
---|
comment:1 Changed 11 years ago by dbevans (David B. Evans)
Cc: | devans removed |
---|---|
Owner: | changed from macports-tickets@… to devans@… |
Status: | new → assigned |
comment:2 Changed 11 years ago by dbevans (David B. Evans)
Comments:
- add missing build dependency gnome-doc-utils
- add missing lib dependency rarian (provides scrollkeeper-update not just at build time but any time port is activated hence lib)
- remove --disable-silent-rules (not supported by this package). Since it doesn't support silent rules, it builds verbosely anyway. To see what is supported, run './configure --help' in the port's work directory.
- 'port -d configure' shows a non-fatal error as follows (need to look at configure and see what command is failing):
... checking for gmsgfmt... /opt/local/bin/msgfmt checking for xgettext... /opt/local/bin/xgettext ./configure: line 13383: ./po/POTFILES.in: No such file or directory checking gnome-doc-utils >= 0.3.2... yes ...
- can't use 'livecheck.type gnome/gnome-with-unstable' because mastersites is not the GNOME source repository. See, for instance, port gegl for a more generic livecheck setup example.
- something I forgot to mention WRT previous ports is that when specifying a dependency on glib2, you should use a path style depspec rather than the port style so that the dependency can be satisfied by either glib2 or glib2-devel. ryandesign, who maintains these ports, has fixed up the previous ones for you (he has an eagle eye for this :-) ). The normal usage is
path:lib/pkgconfig/glib-2.0.pc:glib2
In light of use of gnome-doc-utils, scrollkeeper, no support for --disable-silent-mode, this seems to use an older (GNOME 2) configuration style.
comment:3 follow-up: 5 Changed 11 years ago by juanrgar (Juan R. García Blanco)
Thank you. Please see below my comments:
- Added gnome-doc-utils as build dependency
- Added rarian as lib dependency
- I can see --disable-silent-rules if I run ./configure --help. In fact, if I run ./configure --disable-silent-rules I get no errors. Well, after removing the source directory and uncompressing the sources again, I can see a warning. So removed --disable-silent-rules.
- It seems that po/POTFILES.in, which is requested by GETTEXT macros, is not present in the tarball nor in the repo. I think it would suffice to create po/POTFILES.in. The failing code seems to be introduced by AM_GLIB_GNU_GETTEXT. What should we do?
- Fixed livecheck.
- port:glib2 replaced with path:lib/pkgconfig/glib-2.0.pc:glib2
I still can see the scroll-keeper error. In fact, if I run ${prefix}/bin/scroll-keeper outside macports, I get the same error:
getopt: illegal option -- n getopt: illegal option -- s getopt: illegal option -- c getopt: illegal option -- r getopt: illegal option -- o getopt: illegal option -- l getopt: illegal option -- l getopt: illegal option -- k getopt: illegal option -- e getopt: illegal option -- e getopt: illegal option -- p getopt: illegal option -- e getopt: illegal option -- r getopt: illegal option -- p getopt: illegal option -- d getopt: illegal option -- a getopt: illegal option -- t getopt: illegal option -- e getopt: illegal option -- o Usage: scrollkeeper-update [OPTIONS]
Changed 11 years ago by juanrgar (Juan R. García Blanco)
Attachment: | libgrss-Portfile.2.tgz added |
---|
comment:4 follow-ups: 6 7 Changed 11 years ago by juanrgar (Juan R. García Blanco)
I found why I was getting that scrollkeeper-update error. getopt is a build dependency for rarian. At some point it seems that I decided to remove it, and macports let me do that. And apparently the getopt that comes with OSX (at least with Mavericks) does not work well with radian. Maybe we should set getopt as a lib dependency for rarian.
comment:5 Changed 11 years ago by dbevans (David B. Evans)
Replying to juanrgar@…:
- It seems that po/POTFILES.in, which is requested by GETTEXT macros, is not present in the tarball nor in the repo. I think it would suffice to create po/POTFILES.in. The failing code seems to be introduced by AM_GLIB_GNU_GETTEXT. What should we do?
I even tried running autogen.sh thinking that that would regenerate the missing file but no luck. Directory po exists but it is empty.
I note that this does not stop configure from completing and the build is successful. So I'm thinking that they are in the process of implementing internationalization but haven't completed the task as yet. Obviously something is missing.
I would file a ticket upstream and move on with testing tracker, etc for now. Probably should wait on committing this port until you've had an upstream response.
comment:6 Changed 11 years ago by dbevans (David B. Evans)
Replying to juanrgar@…:
I found why I was getting that scrollkeeper-update error. getopt is a build dependency for rarian. At some point it seems that I decided to remove it, and macports let me do that. And apparently the getopt that comes with OSX (at least with Mavericks) does not work well with radian. Maybe we should set getopt as a lib dependency for rarian.
Well, I notice that the build doesn't install any *.omf files so there's really no point in doing the scroll-keeper update post-activate at all. I'd drop the post-activate phase and remove the dependency on rarian until it can be demonstrated it's necessary (presence of installed *.omf files). Another ticket upstream?
comment:7 Changed 11 years ago by dbevans (David B. Evans)
Replying to juanrgar@…:
I found why I was getting that scrollkeeper-update error. getopt is a build dependency for rarian. At some point it seems that I decided to remove it, and macports let me do that. And apparently the getopt that comes with OSX (at least with Mavericks) does not work well with radian. Maybe we should set getopt as a lib dependency for rarian.
Confirmed, scrollkeeper-update is a bash script that executes getopt during its processing. getopt promoted to lib dep in r118077.
comment:8 Changed 11 years ago by juanrgar (Juan R. García Blanco)
Thank you.
I just sent an email to lbgrss' maintainer letting him know that po/POTFILES.in is missing.
Changed 11 years ago by dbevans (David B. Evans)
Attachment: | patch-POTFILES.in-libgrss.diff added |
---|
Patch to address POTFILES.in issues, dependencies, etc
comment:9 Changed 11 years ago by dbevans (David B. Evans)
Cc: | bob4job@… added |
---|
Per email response from upstream maintainer, patch attached that
- applies upstream patch to update autotools configuration (missing POTFILES.in error remains) and reconfigure using updated autogen.sh
- updates dependencies to match updated configure.ac
- creates po subdir and empty po/POTFILES.in in post-patch phase to make intltoolize happy and silence configuration error.
- drops scrollkeeper related items (no *.omf files are installed).
Test results forwarded to upstream maintainer via email, with renewed request to add a po/POTFILES.in file to project tree in git repository. Upstream maintainer copied here as well.
comment:10 Changed 11 years ago by dbevans (David B. Evans)
Added to GNOME-3/stable test branch in r118203, dependency of tracker.
comment:11 Changed 11 years ago by dbevans (David B. Evans)
Resolution: | → fixed |
---|---|
Status: | assigned → closed |
Committed in r119112 after feedback from upstream maintainer. Upstream git now contains an empty po/POTFILES.in so our workaround for this issue can be removed after the next point release.
Ok, I'll take a look.
Note that when you add someone to the Cc: field in the ticket form you need to use the full email address not just the first part.