Created attachment 427918 [details]
I recently saw a buffer overflow in gmrun-0.9.2
After fixing it, I felt I need to port this to gtk3.
I also fixed some other minor thing in the code.
Created attachment 427920 [details]
Pushed to gitlab:
Maybe you could proxy maintain this if desktop-misc people don't disagree :/
Created attachment 428546 [details]
Updated ebuild to point to freshly set up repository.
Added vcs-snapshot to solve gitlab archive naming, as suggested in #575962 comment 3
Created attachment 429108 [details]
unfortunately, there is still segfault...
$ echo $?
$ gdb gmrun
Reading symbols from gmrun...(no debugging symbols found)...done.
Starting program: /usr/bin/gmrun
[Thread debugging using libthread_db enabled]
Using host libthread_db library "/lib64/libthread_db.so.1".
[New Thread 0x7fffec008700 (LWP 4771)]
[New Thread 0x7fffeb807700 (LWP 4772)]
Thread 1 "gmrun" received signal SIGSEGV, Segmentation fault.
0x00007ffff6ff2ad0 in g_type_check_instance_cast () from /usr/lib64/libgobject-2.0.so.0
#0 0x00007ffff6ff2ad0 in g_type_check_instance_cast () from /usr/lib64/libgobject-2.0.so.0
#1 0x0000555555559c4d in ?? ()
#2 0x00007ffff6307541 in __libc_start_main () from /lib64/libc.so.6
#3 0x000055555555a27a in ?? ()
there is a segmentation fault, when compiling with pie.
For now, I just use a -no-pie ugly workaround.
I hope being able to fix this soon. Unfortunately, the error happens in some macro, so backtrace does not help a lot here. Custom widget may need more changes when migrating to gtk3, but I just did it "raw".
A friend of mine also noticed the same behaviour in evince (under fedora for him), but is not yet able to reproduce the bug.
Created attachment 515316 [details]
Thanks, for your comment, as it incited me to have a look at this issue.
debian bug #857065 pointed the exact same issue with the gtk2 only version of gmrun. They included some fedora patch to fix this.
The fix in this version of the ebuild is pretty similar to what was suggested by the patch.
Created attachment 515318 [details]
I also add a live ebuild (gtk3 deprecated functions removal).
live ebuild 9999 works like a charm! thank you :-)
(In reply to John Blbec from comment #11)
> live ebuild 9999 works like a charm! thank you :-)
Does the 1.0.0 new ebuild version also works ? Here it does
(In reply to Pacho Ramos from comment #3)
> Maybe you could proxy maintain this if desktop-misc people don't disagree :/
Yes, as I already maintain it for myself, i'm not against. Feedback incites me to solve issues, when I just use workaround when only for me.
I also see #640018: some people still use this launcher, and bug can be closed if this bump is integrated.
(In reply to Samuel BAUER from comment #12)
> (In reply to John Blbec from comment #11)
> > live ebuild 9999 works like a charm! thank you :-)
> Does the 1.0.0 new ebuild version also works ? Here it does
1.0.0 ebuild works great as well
Created attachment 516766 [details]
new EAPI defines sysconfdir = etc
fix the 1.0.0 ebuild to install with sysconfir = datadir
it seems there is a problem with a configuration in ~/.gmrunrc which has lesser priority than the main one in /usr/share/gmrun/gmrunrc so user's configuration is ignored until you rename or delete the main one.
Created attachment 517084 [details]
Unlike <1.0.0 user configuration and history are now stored under
Or according your settings for XDG_CONFIG_DIR
1.0.1 removes url and extension handlers in config file, they're now managed through .desktop files and mimes.
Created attachment 517432 [details]
Created attachment 517434 [details]
Created attachment 534874 [details]
Created attachment 538038 [details]
Use tar.bz2 archive instead of tar.gz
Thank you for your contribution. I had a short look on the ebuild.
Here a few ideas:
• Please test the ebuild with repoman full -x https://wiki.gentoo.org/wiki/Repoman
• We always try to use the latest EAPI, please bump to EAPI=7
• do not set KEYWORDS in the 9999 ebuild
• KEYWORDS need ~ after a version bump run ekeyword ~all YOUREBUILD
• please fix the header. (see /usr/portage/skel.ebuild and https://devmanual.gentoo.org/ebuild-writing/eapi/)
• We usually sort the KEYWORDS as ekeywords does. This makes comparison between packages easier. Simply run ekeywords on the ebuild to let it sort.
ewarn "Changes since gmrun-0.9.2:"
ewarn "~/.gmrun_history moved to ~/.config/gmrun_history"
ewarn "~/.gmrunrc moved to ~/.config/gmrunrc"
ewarn "URL and extension handlers are now .desktop files"
# I do not fully understand the following line. Please be more specific, what the
# user needs to know here:
ewarn "You may remove gmrun specific handlers in configuration file"
Created attachment 545716 [details]
repoman seems happy now
(In reply to Jonas Stein from comment #21)
> ewarn "URL and extension handlers are now .desktop files"
> # I do not fully understand the following line. Please be more specific,
> what the
> # user needs to know here:
> ewarn "You may remove gmrun specific handlers in configuration file"
Handlers management is now done via .desktop, so relics from custom handlers should be removed from configuration, I added "so" to the modified ebuild ewarn is easier to understand.
Created attachment 545718 [details]
Created attachment 545720 [details]
Thks for gmrun-1.0.2.ebuild which build and run fine here whereas gmrun-1.0.2 amd stable segfaults here.
I can see this thread started quite a long time ago...
I must have missed something.
Comment on attachment 545720 [details]
@@ -12,4 +12,8 @@
+ <flag name="gtk2">Use gtk+:2 instead of default gtk+:3</flag>
+ <flag name="popt">Use dev-libs/popt for parsing options</flag>