Naive debian packages questions
in [UK Linux]
|
Prev: SELinux preventing game running ?
Next: Ogg in Firefox
From: Kevin Buzzard on 4 Sep 2009 05:20 I have some very naive questions stemming from an annoying problem with fetchmail. I just wanted to run them past this newsgroup to see if I understood the situation correctly. I just upgraded from debian etch to lenny (lenny is now debian stable) and precisely one thing broke: fetchmail. My fetchmail was upgraded to some sort of version of fetchmail 6.3.9, which contains, amongst other things, a little bug where messages of length zero may cause fetchmail to time out when talking to a MSExchange IMAP server. [it also contains a couple of more serious bugs, but my understanding is that these have been fixed in the debian package]. http://developer.berlios.de/bugs/?func=detailbug&bug_id=11980&group_id=1824 is the issue at stake. Now it turns out that I get an awful lot of messages of size zero, because I get an awful lot of spam and some of it turns out to have length zero. So at least once a day, fetchmail is hanging on me and I have to log directly into my mail server via some web interface and delete the offending message. Now I assumed that I could just go to backports.org and get a package for fetchmail 6.3.11, where the problem is fixed. But it looks to me like backports is still dishing out 6.3.9. I was going to wait it out, but this problem is really begining to bug me. So what are my options? I could just download the most recent version of fetchmail and compile and make it myself, which is what I'll do today probably. But is this what debian zealots would advise me to do, or would they advise something else? And how can I find out if and when backports will give me a version of fetchmail I'm happier with? I'm just running these comments past the list to see if anyone can envisage a problem with doing the thing which ten years ago I would have done without thinking. I'm not sure I can remember the last time I typed "make install". Still, I guess the world moves on... Kevin
From: Richard Kettlewell on 4 Sep 2009 06:01 Kevin Buzzard <buzzard(a)ic.ac.remove.uk> writes: > And how can I find out if and when backports will give me a version > of fetchmail I'm happier with? I don't know about backports, but I assume it'll follow unstable. The maintainer is evidently aware but hasn't upgraded yet. http://bugs.debian.org/540288 > I'm just running these comments past the list to see if anyone can > envisage a problem with doing the thing which ten years ago I would > have done without thinking. I'm not sure I can remember the last > time I typed "make install". Still, I guess the world moves on... It ought to work. Another option would be to backport the fix (assuming it's the patch given in the bug report) to the Debian stable source package, build a new .deb file and install that. -- http://www.greenend.org.uk/rjk/
From: Tony van der Hoff on 4 Sep 2009 06:09 Kevin Buzzard wrote: > I have some very naive questions stemming from an annoying problem > with fetchmail. I just wanted to run them past this newsgroup to see > if I understood the situation correctly. > > I just upgraded from debian etch to lenny (lenny is now debian stable) > and precisely one thing broke: fetchmail. My fetchmail was upgraded > to some sort of version of fetchmail 6.3.9, which contains, amongst > other things, a little bug where messages of length zero may cause > fetchmail to time out when talking to a MSExchange IMAP server. > [it also contains a couple of more serious bugs, but my understanding > is that these have been fixed in the debian package]. > > http://developer.berlios.de/bugs/?func=detailbug&bug_id=11980&group_id=1824 > > is the issue at stake. > > Now it turns out that I get an awful lot of messages of size zero, > because I get an awful lot of spam and some of it turns out to have > length zero. So at least once a day, fetchmail is hanging on me and > I have to log directly into my mail server via some web interface > and delete the offending message. > > Now I assumed that I could just go to backports.org and get > a package for fetchmail 6.3.11, where the problem is fixed. > But it looks to me like backports is still dishing out 6.3.9. > I see the version as 6.3.9-rc2.4+lenny, so it is quite possible that the backports project have simply patched 6.3.9 for this bug. > I was going to wait it out, but this problem is really begining > to bug me. So what are my options? I could just download the > most recent version of fetchmail and compile and make it myself, > which is what I'll do today probably. But is this what debian > zealots would advise me to do, or would they advise something else? > That's exactly what I would do/have done (e.g virtualbox in lenny is ancient). Make sure you uninstall the lenny package before you do so, and install your DIY version in /usr/local. > And how can I find out if and when backports will give me a version > of fetchmail I'm happier with? > I'm sure the info's available, but don't know where OTTOMH. > I'm just running these comments past the list to see if anyone can > envisage a problem with doing the thing which ten years ago I would > have done without thinking. I'm not sure I can remember the last time > I typed "make install". Still, I guess the world moves on... > > Kevin
From: Kevin Buzzard on 4 Sep 2009 06:14 For completeness... Tony van der Hoff <tony(a)nospam.vanderhoff.org> wrote: > I see the version as 6.3.9-rc2.4+lenny, so it is quite possible that the > backports project have simply patched 6.3.9 for this bug. ....my understanding is that in fact the backports project patched 6.3.9 for an unrelated but much more serious bug. Thanks for both comments. It hadn't occurred to me that I could make my own deb file. Somehow there are clearly several related ways to solve the problem---I just wanted to do what was generally regarded as "best practice". Kevin
From: Theo Markettos on 11 Sep 2009 11:19
Kevin Buzzard <buzzard(a)ic.ac.delete.uk> wrote: > Thanks for both comments. It hadn't occurred to me that I could make > my own deb file. Somehow there are clearly several related ways to > solve the problem---I just wanted to do what was generally regarded > as "best practice". Sometimes, when faced with this situation, I've grabbed the source package from Debian testing/unstable and tried to munge the control file enough to allow me to build and install it without having to first build a pile of dependent packages first. Effectively I'm backporting the new source package. How often it works depends on how complex the package is an how many dependencies there are. But if the package is already in backports I do wonder if there's a reason why they haven't picked the newer version. Theo |