Hi,
currently the existence of RPATHS is only checked when one does a local rpm build. mock and koji still let it slide. As a result we have these bugs: https://bugzilla.redhat.com/show_bug.cgi?id=703636 https://bugzilla.redhat.com/show_bug.cgi?id=716572 I provided a patch for the first one and even applied for package commit access to apply the fix, yet the maintainers choose to ignore me. Forcing the rpath checker on the build system would at least prevent such problems from happening in the future. I don't want to be pointing fingers, but it seems like a long time problem that some @redhat maintainers happily choose to ignore packaging guidelines [1] without consequences.
Julian
[1] http://fedoraproject.org/wiki/Packaging:Guidelines#Beware_of_Rpath
On Sat, 2011-06-25 at 13:56 +0200, Julian Sikorski wrote:
Hi,
currently the existence of RPATHS is only checked when one does a local rpm build. mock and koji still let it slide. As a result we have these bugs: https://bugzilla.redhat.com/show_bug.cgi?id=703636 https://bugzilla.redhat.com/show_bug.cgi?id=716572 I provided a patch for the first one and even applied for package commit access to apply the fix, yet the maintainers choose to ignore me.
Well, it's fairer to assume the maintainer is simply busy and hasn't got to the bug yet than to assume they're ignoring you.
Note that your patch isn't the best fix; it would be better to use the --disable-rpath configure parameter. I meant to push a build this week which would fix this but didn't get around to it yet...
On Sat, Jun 25, 2011 at 6:58 PM, Adam Williamson awilliam@redhat.com wrote:
On Sat, 2011-06-25 at 13:56 +0200, Julian Sikorski wrote:
Hi,
currently the existence of RPATHS is only checked when one does a local rpm build. mock and koji still let it slide. As a result we have these bugs: https://bugzilla.redhat.com/show_bug.cgi?id=703636 https://bugzilla.redhat.com/show_bug.cgi?id=716572 I provided a patch for the first one and even applied for package commit access to apply the fix, yet the maintainers choose to ignore me.
Well, it's fairer to assume the maintainer is simply busy and hasn't got to the bug yet than to assume they're ignoring you.
Note that your patch isn't the best fix; it would be better to use the --disable-rpath configure parameter. I meant to push a build this week which would fix this but didn't get around to it yet...
https://admin.fedoraproject.org/updates/gnome-shell-3.0.2-4.fc15
On Sat, 25 Jun 2011 19:24:47 +0200 drago01 drago01@gmail.com wrote:
On Sat, Jun 25, 2011 at 6:58 PM, Adam Williamson awilliam@redhat.com wrote:
On Sat, 2011-06-25 at 13:56 +0200, Julian Sikorski wrote:
Hi,
currently the existence of RPATHS is only checked when one does a local rpm build. mock and koji still let it slide. As a result we have these bugs: https://bugzilla.redhat.com/show_bug.cgi?id=703636 https://bugzilla.redhat.com/show_bug.cgi?id=716572 I provided a patch for the first one and even applied for package commit access to apply the fix, yet the maintainers choose to ignore me.
Well, it's fairer to assume the maintainer is simply busy and hasn't got to the bug yet than to assume they're ignoring you.
Note that your patch isn't the best fix; it would be better to use the --disable-rpath configure parameter. I meant to push a build this week which would fix this but didn't get around to it yet...
https://admin.fedoraproject.org/updates/gnome-shell-3.0.2-4.fc15
I'll note here that _removing_ rpath completely breaks gnome-shell.
Likely this is something that should be addressed upstream before removing rpath is possible in this case at least.
kevin
Kevin Fenzi kevin@scrye.com writes:
On Sat, 2011-06-25 at 13:56 +0200, Julian Sikorski wrote:
currently the existence of RPATHS is only checked when one does a local rpm build.
I'll note here that _removing_ rpath completely breaks gnome-shell.
It would also break a large number of perl-dependent packages, since we still have libperl.so not being installed into /usr/lib. I don't think that having the build system enforce a prohibition would be a good idea at all.
regards, tom lane
On Sat, 2011-06-25 at 19:24 +0200, drago01 wrote:
Note that your patch isn't the best fix; it would be better to use the --disable-rpath configure parameter. I meant to push a build this week which would fix this but didn't get around to it yet...
https://admin.fedoraproject.org/updates/gnome-shell-3.0.2-4.fc15
I was actually talking about cheese, not gnome-shell; sorry for not clarifying.
W dniu 25.06.2011 18:58, Adam Williamson pisze:
On Sat, 2011-06-25 at 13:56 +0200, Julian Sikorski wrote:
Hi,
currently the existence of RPATHS is only checked when one does a local rpm build. mock and koji still let it slide. As a result we have these bugs: https://bugzilla.redhat.com/show_bug.cgi?id=703636 https://bugzilla.redhat.com/show_bug.cgi?id=716572 I provided a patch for the first one and even applied for package commit access to apply the fix, yet the maintainers choose to ignore me.
Well, it's fairer to assume the maintainer is simply busy and hasn't got to the bug yet than to assume they're ignoring you.
Note that your patch isn't the best fix; it would be better to use the --disable-rpath configure parameter. I meant to push a build this week which would fix this but didn't get around to it yet...
My patch is not the best fix, true. The problem is that --disable-rpath does not work as advertised and using the sed command from the Fedora wiki prevents cheese from building at all. It is all written down in the bug.
Julian
On Sat, 2011-06-25 at 22:06 +0200, Julian Sikorski wrote:
W dniu 25.06.2011 18:58, Adam Williamson pisze:
On Sat, 2011-06-25 at 13:56 +0200, Julian Sikorski wrote:
Hi,
currently the existence of RPATHS is only checked when one does a local rpm build. mock and koji still let it slide. As a result we have these bugs: https://bugzilla.redhat.com/show_bug.cgi?id=703636 https://bugzilla.redhat.com/show_bug.cgi?id=716572 I provided a patch for the first one and even applied for package commit access to apply the fix, yet the maintainers choose to ignore me.
Well, it's fairer to assume the maintainer is simply busy and hasn't got to the bug yet than to assume they're ignoring you.
Note that your patch isn't the best fix; it would be better to use the --disable-rpath configure parameter. I meant to push a build this week which would fix this but didn't get around to it yet...
My patch is not the best fix, true. The problem is that --disable-rpath does not work as advertised and using the sed command from the Fedora wiki prevents cheese from building at all. It is all written down in the bug.
Last I read, --disable-rpath worked if you did an autoreconf too? That would still be the most preferable option, I believe...
W dniu 25.06.2011 23:43, Adam Williamson pisze:
On Sat, 2011-06-25 at 22:06 +0200, Julian Sikorski wrote:
W dniu 25.06.2011 18:58, Adam Williamson pisze:
On Sat, 2011-06-25 at 13:56 +0200, Julian Sikorski wrote:
Hi,
currently the existence of RPATHS is only checked when one does a local rpm build. mock and koji still let it slide. As a result we have these bugs: https://bugzilla.redhat.com/show_bug.cgi?id=703636 https://bugzilla.redhat.com/show_bug.cgi?id=716572 I provided a patch for the first one and even applied for package commit access to apply the fix, yet the maintainers choose to ignore me.
Well, it's fairer to assume the maintainer is simply busy and hasn't got to the bug yet than to assume they're ignoring you.
Note that your patch isn't the best fix; it would be better to use the --disable-rpath configure parameter. I meant to push a build this week which would fix this but didn't get around to it yet...
My patch is not the best fix, true. The problem is that --disable-rpath does not work as advertised and using the sed command from the Fedora wiki prevents cheese from building at all. It is all written down in the bug.
Last I read, --disable-rpath worked if you did an autoreconf too? That would still be the most preferable option, I believe...
I just checked again. - autoreconf does not seem to make a difference - ./autogen.sh is enough to remove rpaths, --disable-rpath is not needed anymore. If running ./autogen.sh is preferred to chrpath, then I think this should be documented in the wiki. I had exactly the same issue with gnome-chemistry-utils long time ago and I ended up using chrpath.
Julian
On Sun, 2011-06-26 at 08:43 +0200, Julian Sikorski wrote:
I just checked again.
- autoreconf does not seem to make a difference
- ./autogen.sh is enough to remove rpaths, --disable-rpath is not needed
anymore. If running ./autogen.sh is preferred to chrpath, then I think this should be documented in the wiki. I had exactly the same issue with gnome-chemistry-utils long time ago and I ended up using chrpath.
I'm not sure, but I'd guess that chrpath is correct in this case, as it looks like autogen.sh removing rpaths is more or less an accident/coincidence...so probably chrpath is better as it's at least intended to have the effect we desire. So you were right all along, sorry =) I'll apply your patch and push a build soon, if the desktop team does nothing...