Locally created mailboxes not synchronized to IMAP

[Vladimir Marek]

> > Would you know how to reproduce that?
> 
> See below. Summary: I see 2 things that could be improved a) allow empty
> foldernames through nametrans and b) allow Maildirs to start with '/'.
>  
> You are trying to map this with 
> re.sub('^outside/?', '', foldername)?

Well, in previous mail you said:

========================================================================
> > -        for dirname in os.listdir(toppath) + ['.']:
> > +        for dirname in os.listdir(toppath):
> >              self.debug("  *** top of loop")
> 
> Careful here! I just had to fix this code as I had it broken before. We
> do want to examine all directories *including* the current one. What is
> the reason for removing this?
> You are omitting the top directory now!
========================================================================

I'm trying to make test case for it, so that I will know when I broke
it :)


> >  * re.sub('^outside/?', '.', foldername)
> > maildir/new
> > maildir/cur
> > maildir/cur/U=1,FMD5=5058f1af8388633f609cadb75a75dc9d:2,S
> > maildir/tmp
> > maildir/.inside
> > maildir/.inside/tmp
> > maildir/.inside/cur
> > maildir/.inside/new
> > maildir/.inside/new/U=5,FMD5=73ee1943692f76db315f5374b153650b:2,
> > ....
> >   File "/builds/vm156888/offlineimap/offlineimap/repository/Maildir.py", line 85, in makefolder                                                                                   
> >     assert not foldername.startswith('/'), "Folder names may not begin with /"                                                                                                    
> > AssertionError: Folder names may not begin with /  
> 
> Interesting why you would have a folder name here that starts with
> '/'. Can you debug print the folder name it attempts to create here?
> Is it the remote folder '/' ?

Good questions :)

> 
> In any case, this one seems something that can be improved!
> I am not sure why we shouldn't allow Maildir folder names starting with
> '/'. This is an old check and I am sure we can deal with that
> somehow. We will get into this situation only if:
> Maildir.sep=/ and foldername is '/*'. What would happen if we simply
> stripped any leading '/' from Maildir foldernames?
> 
> '/' would become '' and would be represented by the folders:
>  maildir/cur
>  maildir/new

I might look into it, as I will be testing my folder synchronization
changes.


> More reasons to have a comprehensive test suite that tests a wide range
> of configuration situations.

Exactly

-- 
	Vlad