On Sat, 04 Jul 2009, 12:41 +0200, Stephan Bosch wrote:
John Marshall schreef:
On Tue, 23 Jun 2009, 18:29 +0200, Stephan Bosch wrote:
John Marshall wrote:
--< snip: 169 lines removed >-- # This is a sieve script file fuNO "Unknown error"
There's our problem. There's only one location in the sources where this most informative error can occur :) . Apparently, Timo has fixed an old bug in the Dovecot stream implementation. This bugfix triggers a problem in ManageSieve that uses an older section of code of the imap daemon. Timo provided the solution and I've committed the fix:
http://hg.rename-it.nl/dovecot-1.1-managesieve/rev/75c9769339b8
Please apply this to your managesieve source and check whether the problem is gone. I'll make a new release when all is ok.
Thank you Stephan,
Was the patch for cmd-getscript.c the only patch? I applied that patch to cmd-getscript.c and rebuilt. That eliminated the informative error message but still does not return the entire script.
GETSCRIPT "test_9000" {9223} # This is a sieve script file full of comments --< snip: 169 lines removed >-- # This is a sieve script file fu OK "Getscript completed."
I was hoping that this uninformative error was now spurious, meaning that my 'fix' indeed only got rid of the error message. However, as it seems, the stream is still truncated. I rebuilt the truncation check, now with a proper error and disconnect (because the protocol state is invalid then).
However, regarding the core of this problem, I am unable to reproduce this at my end without access to a FreeBSD system. Could you (or some other FreeBSD user) try to dig a little deeper to find out exactly why the stream ends prematurely? Alternatively, I would have to install FreeBSD somewhere, which will take more time.
I happy to help but I don't know where/how to start. Are you able to provide me with some digging hints?
Thank you.
-- John Marshall