[Pidgin] #1349: Failure to detect dbus services directory
Pidgin
trac at pidgin.im
Mon May 28 03:39:58 EDT 2007
#1349: Failure to detect dbus services directory
--------------------------+-------------------------------------------------
Reporter: jmmv | Type: defect
Status: new | Priority: minor
Component: pidgin (gtk) | Version: 2.0.1
Keywords: | Pending: 0
--------------------------+-------------------------------------------------
---- begin of copy/paste from gaim's bug tracker ----
NetBSD installs third-party software in the /usr/pkg hierarchy. Among
these applications is dbus, which has its services directory under
/usr/pkg/share/dbus-1/services.
Unfortunately, gaim's configure does not detect that directory properly,
although a quick inspection to the current code shows that it should.
These are the directories in which gaim looks:
serviceprefixes="$datadir $libdir /usr/share /usr/local/share"
According to this, $datadir should expand to /usr/pkg/share, thus making
the following loop find the directory. But this is false. Neither of
$datadir nor $libdir are expected to be expanded from the configure
script. If you try to do so, they expand to ${prefix}/share and
${prefix}/lib respectively. I mean, the ${prefix} part is left
unexpanded. As you can imagine, this makes the future test fail because
there is no directory named '${prefix}...'.
One could think that a correct solution could be to add an extra 'eval' to
the affected code so that it expanded the variables one more time. This
is incorrect (and difficult to achieve, since nothing warrants that there
is a single extra level of indirection). The autoconf documentation (see
the info file) clearly states that they must only be expanded from
Makefiles.
I've fixed this by replacing $datadir with $prefix/share and $libdir with
$prefix/lib. This might not be 100% perfect, but after all configure is
trying to find something on its own. It is not required to find a match,
even more if you have a nonstandard layout (e.g., changing where the
datadir is). So, in my oppinion, this is safe to do.
---- end of copy/paste from gaim's bug tracker ----
OK. The way you have managed this site migration has been extremely
annoying: renaming a project and moving it to another site DOES NOT FIX
BUGS. You have made me lose a lot of time figuring out how to fetch
pidgin's code, reapplying the patch, registering an accout here and
resubmitting the problem report. Sincerely, I don't know why I bothered
as I don't use pidgin any more. I bet you have lost a lot of good problem
reports along the migration, as the original submitters will not resend
them. (I was already familiar with Monotone, but many of the original but
reporters are probably not.)
Not to mention that the patch to this bug is trivial and that it could
have been applied a long time ago. Or you could checked if it was still
applicable by minor, minor inspection to the code. But oh well, looks
like NetBSD-related portability problems are not important in today's
Linux/i386-only world.
--
Ticket URL: <http://developer.pidgin.im/ticket/1349>
Pidgin <http://pidgin.im>
Pidgin
More information about the Tracker
mailing list