Bug 903667 - the "offlineimap-6.5.2.1-3.el6" update breaks existing .offlineimaprc files on Red Hat *ENTERPRISE* Linux
Summary: the "offlineimap-6.5.2.1-3.el6" update breaks existing .offlineimaprc files o...
Keywords:
Status: CLOSED NOTABUG
Alias: None
Product: Fedora EPEL
Classification: Fedora
Component: offlineimap
Version: el6
Hardware: Unspecified
OS: Unspecified
low
high
Target Milestone: ---
Assignee: Paolo Bonzini
QA Contact: Fedora Extras Quality Assurance
URL:
Whiteboard:
Depends On:
Blocks:
TreeView+ depends on / blocked
 
Reported: 2013-01-24 14:38 UTC by Laszlo Ersek
Modified: 2013-01-25 10:37 UTC (History)
1 user (show)

Fixed In Version:
Clone Of:
Environment:
Last Closed: 2013-01-25 10:37:27 UTC
Type: Bug
Embargoed:


Attachments (Terms of Use)

Description Laszlo Ersek 2013-01-24 14:38:36 UTC
**** Description of problem:
In today's "yum update", offlineimap-6.3.4-1.el6 was replaced on my RHEL-6 system, with offlineimap-6.5.2.1-3.el6. Consequently, my ~/.offlineimaprc file stopped working:

----------v----------
[general]
accounts = RedHat
ui = TTY.TTYUI

[Account RedHat]
localrepository = Local
remoterepository = Zimbra
maxconnections = 5

[Repository Local]
type = Maildir
localfolders = ~/Maildir
existsok = yes

[Repository Zimbra]
type = IMAP
remotehost = mail.corp.redhat.com
remoteuser = lersek
ssl = yes

# For Dovecot to see the folders right I want them starting with a dot and the remote 'INBOX' as the toplevel Maildir
nametrans = lambda foldername: re.sub('^\.INBOX$' , '.', re.sub('^', '.' , foldername))

folderfilter = lambda foldername: not re.search('Archives/backup-in-mailman/', foldername)
----------^----------

**** Version-Release number of selected component (if applicable):
offlineimap-6.5.2.1-3.el6

**** How reproducible:
100%

**** Steps to Reproduce:
1. update from offlineimap-6.3.4-1.el6 to offlineimap-6.5.2.1-3.el6
2. try to sync
  
**** Actual results:
offlineimap refuses to synchronize, with the following message:

----------v----------
 ERROR: INFINITE FOLDER CREATION DETECTED! Folder '.Archives.backup-in-mailman' (repository 'Local') would be created as folder '/Archives/backup-in-mailman' (repository 'Zimbra'). The latter becomes '..Archives.backup-in-mailman' in return, leading to infinite folder creation cycles.
 SOLUTION: 1) Do set your nametrans rules on both repositories so they lead to identical names if applied back and forth. 2) Use folderfilter settings on a repository to prevent some folders from being created on the other side.
----------^----------

**** Expected results:
The update is unnoticeable to the user, offlineimap keeps working.

**** Additional info:
This update is a *grave* violation of EPEL guidelines.

From <http://fedoraproject.org/wiki/EPEL/GuidelinesAndPolicies#Some_examples_of_what_package_updates_that_are_fine_or_not>:

----------v----------
A yet again little bit bigger minor version updates

Let's assume package foo is shipped in EPEL 5.0 as version 1.0.1; upstream developers now ship 1.4.0; the ABI is compatible to 1.0.1, but the config files need manual adjustments

- build for the stable branch is normally not acceptable; a backport should be strongly considered if there are any serious bugs that must be fixed 
----------^----------

This section applies because:
- offlineimap-6.3.4-1.el6 
- offlineimap-6.5.2.1-3.el6
                ^
                |
minor release bump requiring manual config file changes 

The only bug that got fixed in this new EPEL-6 release is bug 835688. The patch attached to that bug is small; it should have been backported to 6.3.4.

For the time I added

----------v----------
[epel]
# ...
exclude = offlineimap
----------^----------

to "/etc/yum.repos.d/epel.repo".

Please fix ASAP. I'm not interested in reworking my config file; I'm on *RHEL* and not Fedora because regressions kill my productivity.

Comment 1 Laszlo Ersek 2013-01-24 14:40:09 UTC
Actually I can live with the workaround; setting Prio to Low and lowering Severity to High.

Comment 2 Paolo Bonzini 2013-01-25 10:37:27 UTC
To me it looks like the complaint is valid and the configuration file was wrong to begin with.


Note You need to log in before you can comment on or make changes to this bug.