On Thu, Jan 15, 2009 at 03:23:55AM +0200, Eddy Petrișor <eddy.petri...@gmail.com> was heard to say: > Daniel Burrows a scris: > > On Tue, Jan 13, 2009 at 08:28:21PM +0200, Eddy Petrișor > > <eddy.petri...@gmail.com> was heard to say: > >> I just upgraded this system yesterday to Debian Lenny from Debian Etch > >> and since the upgrade I have seen that aptitude's TUI interface blocks > >> after any set of operations that is done with dpkg. When the operation > >> finishes the TUI is unusable and you need to ctrl+C the interface to > >> get the prompt. > > > > What architecture are you running on? Is this a single or a multi-core > > Err, I meant to add that info, too, sorry about that. > > Arch is i386, dual core. > > > machine? Just for kicks, if you install aptitude-dbg and valgrind, then > > run "valgrind --log-file=/tmp/aptitude.grind", can you still reproduce > > the problem? (warning: this will be very slow) > > I'll try this tomorrow when I get at work (that's where the machine is).
Something that would be even better if you can manage it would be to use "hg bisect". This will be time-consuming, but has the potential to track down exactly what's happening, or at least what change triggered it. The one time I was able to (very briefly) reproduce this I tried running valgrind and gdb, but without getting anywhere. If this is too much work for you that's OK -- these instructions might help other people track it down. Heck, I can even post them on my blog. Can't hurt. (why didn't I do this before? I only thought of using "bisect" while I was walking to work this morning. D'oh!!) I don't know what version previously worked for you, but etch has 0.4.4-4, so let's try that. # aptitude $ hg clone http://hg.debian.org/hg/aptitude/head aptitude $ cd aptitude $ hg bisect --bad $ hg bisect --good 0.4.4 Testing changeset 1105:083027d51594 (1098 changesets remaining, ~10 tests) 256 files updated, 0 files merged, 78 files removed, 0 files unresolved Then repeat these steps: $ ./autogen.sh && ./configure && make # ./aptitude (try to reproduce the problem) After each attempt, if you can reproduce it, tell Mercurial that the version is buggy by runnning: $ hg bisect --bad If not, tell Mercurial that it isn't buggy by running: $ hg bisect --good If you can't compile the program, or if it's too broken for you to test it, run: $ hg bisect --skip It should take around 10 tries, more if you're lucky. Then it'll print a changeset out: The first bad revision is: changeset: 1446:d2da6ae09bd3 parent: 1436:9db554d44e20 parent: 1435:78895fdf40d7 user: Christian Perrier <bubu...@debian.org> date: Sat Apr 12 09:00:07 2008 +0200 summary: Merging changes to fr.po Send me the changeset information Mercurial gives you. Daniel -- To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org
