Re: Re: Lintian's warnings : ldconfig call is useless and unsafe...
Hi Andreas
I noticed that you commited today further changes which I did not yet
checked - the build takes some time as you know.
I pushed the commit 7baa5930725add6feb4517e0f6a605762af6e870 last night, which
fix some of the following points. But I will come back on them now :
When I builded commit
ID aebf1274866f7e5417efdd4ee8aa3fbdf50e6dc1 and using the last lintian
version from unstable (which might be more picky than some other
version) I've got:
E: fw4spl source: source-is-missing Apps/VRRender/doc/source/_static/jquery.js
-> Seems to remain
I thought it was fixed by the commit b713ebad050ea798e8e57710ae5b212888a48d32, in
which I exluded the Apps/VRRender/doc directory from d/copyright. I guess I
misunderstood your answer about it. Is is not enough ? Must I to remove them
from the upstream tarball ? I'd rather don't do that because I guess that if this
files are there, developers had certainly a good reason. But I'm not sure.
W: fw4spl source: changelog-should-mention-nmu
W: fw4spl source: source-nmu-has-incorrect-version-number 0.9.2-1
-> Seems you fixed both in last commit
Yes I do. It was caused by different email addresses and names between d/control
and d/changelog.
I: fw4spl source: quilt-patch-missing-description remove_fwItkIOTest.patch
I: fw4spl source: quilt-patch-missing-description set_media_as_bundle.patch
I: fw4spl source: quilt-patch-missing-description add_easy_launcher.patch
I: fw4spl source: quilt-patch-missing-description fix_unit_tests.patch
-> Please add DEP3 descriptions (if in doubt what this mean you might like
to use `lintian -i ...`).
I work on it today.
W: fw4spl source: timewarp-standards-version (2014-07-16 < 2014-09-17)
-> Seems to be fixed.
Yes it is.
W: fw4spl: binary-without-manpage usr/bin/launcher
W: fw4spl: binary-without-manpage usr/bin/launcher-0.1
-> Would be nice to have but we could live with this for the moment.
However, what I'm more concerned about is the very generic name of the
binary. I wonder whether /usr/bin/fw4spl or /usr/bin/fw4spl-launcher
would be more apropriate and this also should be suggested to upstream.
Short and simple manpages have been added by the last commit. About the binary name, I agree
with you, I work in it the afternoon.
so there are some more things left to do. Specifically the first one is
an error. I guess this is connected to your recent question about
recreating the upstream tarball. Can you confirm that you worked on
this?
Yes it's connected (see above).
I don't understand why I get these errors. ldconfig is used to register the shared
libraries. For this warning, I think I'll override the lintian issue as it was done
for a lot of other packages [1]
Here is my postinst script :
if [ "$1" = "configure" ]; then
... && ldconfig
fi
But about the unsafe call to ldconfig, I thought fix it by adding this in postrm :
if [ "$1" = "remove" ]; then
... && ldconfig
fi
I've also try to do :
test "$1" = "remove" || exit 0
... && ldconfig
But it doesn't fix the issue, as expected... Is it not safe ? And why is it
useless ?
I guess the call is useless since the dynamic libraries are in a subdir
of /usr/lib. As far as I know ldconfig deals with this automatically.
I'm not sure whether your new means in postinst / postrm is fixing the
issue. I think you can safely remove these files from the packaging.
It's especially because ldconfig doesn't deal with my subdirectory that I try to edit
this files. In the last files committed, I try this :
#!/bin/sh
set -e
if [ "$1" = "configure" ]; then
sh -c "echo '/usr/lib/fw4spl' > /etc/ld.so.conf.d/fw4spl.conf"
fi
#DEBHELPER#
exit 0
(I had this warning : maintainer-script-lacks-debhelper-token)
And it's " #DEBHELPER# " which add the call to ldconfig...
Since I'm a bit in a hurry I'd recommend to ask on
debian-mentors@lists.debian.org where this kind of question is easily
handled.
Yes I'll ask them about it.
Thank you !
Best regards,
Corentin Desfarges
Reply to: