ANNOUNCEMENT: Live Wireshark University & Allegro Packets online APAC Wireshark Training Session
April 17th, 2024 | 14:30-16:00 SGT (UTC+8) | Online

Wireshark-dev: Re: [Wireshark-dev] building a Gtk2 RPM (Was: QT_MIN_VERSION)

From: Jeff Morriss <jeff.morriss.ws@xxxxxxxxx>
Date: Wed, 13 Aug 2014 15:42:36 -0400
On 08/12/14 18:51, Christopher Maynard wrote:
Jeff Morriss <jeff.morriss.ws@...> writes:

That's odd; I just tried it (starting from the wireshark-1.12.0 source
tarball[1]) and did not have any problems.  I wonder if your
wireshark.spec file isn't being regenerated (from wireshark.spec.in)?
You could try comparing the two or just remove wireshark.spec to force
it to be built again.

[1] cd /tmp && tar xjf /path/to/wireshark-1.12.0.tar.bz2 && cd
wireshark-1.12.0 && ./configure --with-gtk2 && make rpm-package

Thanks for retesting.  The wireshark.spec file was definitely being regenerated.

I noticed above that you did not run ./autogen.sh.  I tried once again, this
time *without* running autogen.sh just as you've shown, but unfortunately
the results were the same.

Source tarballs don't need ./autogen.sh run on them--that's part of the magic of "make dist" (and part of the argument against distributing source tarballs with tools/git-export-release.sh).

Anyway, I had been running this test on Fedora (20) but since I didn't want to accuse you of operator error ;-) I decided to try it on RHEL 6. Sure enough, I saw the problem. And in fact at the moment I'm not sure why it works on Fedora--I'll need to dig a bit deeper. But anyway I think I know the fix--I'll commit that once I either have a better understanding of why it works on Fedora or at least I prove it works in both distros.