I just started seeing this problem as well, and the hack didn't work, nor did upgrading to the newest RC Release (0.2.1).
It seems to me, although this could be a red herring, that it began after my inbox grew past 900 unread messages.
Which, for me, is normal.
When I used SquirrelMail - I had filters and the like setup to move my subscribed lists and daily incoming mail to numerous other folders (like .. well .. most email clients let you do [thunderbird, squirrelmail, outlook, xmail, elm, etc etc]).
Now that I don't have that ability - my inbox grows quite fast, and within a week easily expands past 1000 messages.
Is it possible the AJAX message is reaching a maximum length and bombing?
There are no errors in "logs/errors"
EDIT: More Information
I turned the debug console on, and now when the failure happens I can see what it does, but not why it does it.
When I click the "Inbox" - the return from the AJAX call is all wrong - it is returning the index page from my default host NOT the configured virtual host this application responds to. I can replicate it by taking the HTTP Request that I see in the console and applying it to my webmail URL. I tested this by reconfiguring my index.html on my default host with a simple page - which from the debug below we can see is being returned by AJAX.
--------------------------------------
HTTP request: ./?_task=mail&_action=list&_mbox=INBOX&_page=1&_refresh=1&_remote=1
--------------------------------------
<center><link rel=stylesheet href=http://credoaction.com/widgets/widgetstyle.css
type=text/css media=screen /><script language=javascript
src=http://credoaction.com/widgets/widget.js></script>
<script language=javascript>WriteBushCountdown()</script>
But, the very next thing I do is click another folder with far less information in it, and the response looks more normal:
--------------------------------------
HTTP request: ./?_task=mail&_action=list&_mbox=ham&_page=1&_refresh=1&_remote=1
--------------------------------------
/** ajax response [30/Apr/2009 01:21:33 -0400] **/
this.set_env({mailbox:'ham',quota:1,delimiter:'.',trash_mailbox:'Trash',drafts_mailbox:'Drafts',junk_mailbox:'spam',messagecount:1,pagecount:1});
this.set_pagetitle('**Scrubbed** :: ham');
this.set_rowcount('Messages 1 to 1 of 1');
this.set_mailboxname('ham');
this.set_message_coltypes(['from','subject','date','size','flag']);
this.add_message_row('1',{from:'Amy Goodman<\/span>',subject:'[SPAM] Democracy Now! Needs You<\/a>',date:'12.12.2007 20:22',size:'100 KB'},{deleted:0,unread:0,replied:0,forwarded:0,flagged:0},1,0);
this.set_unread_count('ham',0,0);
This tells me that either 1 of 2 things is going on. Either Apache forgets how to forward requests to a virtual host when RoundCube has a large list to transfer (unlikely) or -- When RoundCube tries to refresh a very large list - it fails, and the result is a forward to something that does NOT correspond to my virtual host (which is only
http://mail.some_host.com/) causing the refresh to fail.
Please advise.
P