sieve-connect doesn't like to be called without parameters
Affects | Status | Importance | Assigned to | Milestone | |
---|---|---|---|---|---|
sieve-connect (Ubuntu) |
Fix Released
|
Undecided
|
Phil Pennock |
Bug Description
Binary package hint: sieve-connect
If you call sieve-connect without parameters it will go into an endless loop spitting out
Use of uninitialized value $fam_listen in numeric ne (!=) at /usr/share/
all over your screen. It should do something reasonable, like calling --help.
I also tried 0.73-1 and the behaviour persists. There's an updated version 0.77 at http://
ProblemType: Bug
DistroRelease: Ubuntu 10.10
Package: sieve-connect 0.69-1
ProcVersionSign
Uname: Linux 2.6.35-25-generic i686
Architecture: i386
Date: Thu Feb 24 12:47:11 2011
InstallationMedia: Kubuntu 10.04 LTS "Lucid Lynx" - Release i386 (20100427)
PackageArchitec
ProcEnviron:
LANGUAGE=
PATH=(custom, user)
LANG=en_US.UTF-8
SHELL=/bin/bash
SourcePackage: sieve-connect
I'm the author of this code and only just found this bug-report at Ubuntu. What should happen is something like this:
### ./ because I just checked out rev 69 of the code :sieve( 4190)> failed.
% ./sieve-connect.pl
Connection to <[localhost]
That's because if you set $IMAP_SERVER in the environment, then when invoked with no arguments it connects straight in, so the "just connect to best derived destination" is the default behaviour.
I can't reproduce this upstream (non-Ubuntu) and the error message is inside one of the dependent Perl modules, referencing a variable that's not part of my code, so I think that this is a bug in one of the Perl libraries, at a particular point in time. "sieve-connect --debug --version" will tell you the versions of the main libraries it depends upon; for me, I see:
Module IO::Socket::INET6 Version 2.69
On an Oneiric VM, I see:
Module IO::Socket::INET6 Version 2.65
and the correct error message, no loop.
Looking through http:// cpansearch. perl.org/ src/SHLOMIF/ IO-Socket- INET6-2. 69/ChangeLog I don't see anything which appears to be relevant.
So for now I'm going to pass on making any changes, since it looks like a bug in a version of the code I can't identify, already fixed in more recent versions of that code, and where I've no idea what I might be able to do to work around it for the bad library versions.