Comments on: Comparing Debian and Fedora

By: jef

jef — Tue, 11 Sep 2007 04:37:31 +0000

“jef: IMHO the updates-testing (if you choose to use it – and I don’t) makes it closer to Debian/Unstable.”

That doesn’t make any sense to me. Updates-testing does not get all the updates that rawhide gets. Updates-testing only gets updates that are meant for the stable releases. You probably need to look at the Fedora structure closer.

By: etbe

etbe — Mon, 10 Sep 2007 05:08:30 +0000

jef: IMHO the updates-testing (if you choose to use it – and I don’t) makes it closer to Debian/Unstable.

By: jef

jef — Sun, 09 Sep 2007 20:32:13 +0000

“Fedora releases (previously known as “Fedora Core” and now merely as “Fedora”) can be compared to Debian/Testing.”

Umm. What about Fedora updates-testing then? There goes your analogy.

By: the debian user » Blog Archive » Comparing Debian and Fedora

the debian user » Blog Archive » Comparing Debian and Fedora — Sun, 09 Sep 2007 14:51:00 +0000

[...] and SELinux developer Russell Coker did a nice comparison of Debian and Fedora. I’m often asked about that, even from companies, so if you need this to compare the [...]

By: Adrian Bunk

Adrian Bunk — Sun, 09 Sep 2007 12:00:16 +0000

Regarding the dependency requirement in Debian testing: I remember one point in time when testing contained versions of udev and the ALSA packages that couldn’t be installed together (one package had a versioned conflict with the other) – according to the logic of the testing scripts this is perfectly OK…

And there’s a difference between a coordinated update of some packages and the semi-random daily testing updates – consider e.g. that a bug “new version segfaults every 15 minutes” is usually not considered to be an RC in Debian by your release managers and therefore nothing that would prevent a package from entering testing.

Besides this, the pure number of updates in testing makes it quite likely that you’ll sooner or later run into some regression – and even the testing security updates you mention both logically (no fix if fixed packages are already in testing) and technically (package dependencies) require testing users to regularly update their testing installation.

Another funny effect of testing is (unless this has been changed) that if it takes 2 days until a package has been autobuilt on some architecture, this package might enter testing (consider urgency=high) without ever giving anyone the chance to test the package on this architecture. And that’s not a theoretical issue, when I was a Debian maintainer I once broke util-linux on powerpc this way.