Re: Considering reverting imaplib2.py and IDLE support - need feedback

[Mike Dawson <mike@xxxxxxxxxxxxxxxxx>]

Hi,

Well if there is anything that can be done to leave it as an option then 
that would be great.   I have considered doing the loop - the thing is 
if someone for example changes their password whilst you are syncing 
other accounts then offlineimap would crash even before someone else's 
attachment was downloaded for example.  So I definitely need to do 
something for trying to isolate stuff a bit... 

The strange thing I saw was:
http://www.cs.usyd.edu.au/~piers/python/imaplib.html

Quote from end of first paragraph
" (|imaplib2| can be substituted for |imaplib| in existing clients with 
no changes in the code.) "

I'm not familiar with the details of imaplib vs imaplib2 though...

Regards,

-Mike

Marc MERLIN wrote:
> On Mon, Aug 10, 2009 at 09:34:44AM -0500, John Goerzen wrote:
>   
>> Mike Dawson wrote:
>>     
>>> I think idle support is a really important thing to have and was 
>>> something that mail addicts really appreciate - I mean having to wait 
>>> for the sync to get your spam makes some people twitch.
>>>       
>> <grin>
>>
>> Of course, I agree...  the problem is that the existing code that
>> provides it required a switch to a library that has introduced stability
>> problems.  I think I am going to have to revert it for now.  I just got
>> yet another bug report about it.  I think imaplib2 is not ready for
>> prime time yet.
>>     
>
> For what it's worth, I wrapped offlineimap in a while loop with a quick
> shell script to detect hangs and kill it. 
> I'm not saying that it's a "good" solution, but for me IDLE support is
> required to be able to get mail at all, so it's worth it.
> With my while loop, things work great, even if it's a bit ugly.
>
> Would there be a way to either
> 1) add a built in watchdog in offlineimap to restart de-wedge/restart it as
> required while imaplib2 gets fixed?
> 2) make imaplib2/IDLE a compile option for those who really need it?
> (yes, I know, right now the code doesn't allow for easily making it a
> compile option, but it's a suggestion to the original patch author).
>
> #2 would allow for distros to build offlineimap and offlineimap-idle and
> people use the the one that works for them or the one they need.
> Hopefully eventually offlineimap-idle would become the default again.
>
> Marc
>