[Dovecot] sieve: deleteheader sometimes fails [bug]
Hi,
We're using Dovecot 2.1.16 with Pigeonhole 0.3.4. Mail is delivered by exim4 to the dovecot LDA.
Sometimes users are unable to delete headers. I've narrowed the problem down to preceding the deleteheader command with a query against the body.
A simple example: """ require ["body", "editheader", "fileinto"]; if body :contains "!TEST!" { fileinto "INBOX.Trash"; stop; } deleteheader "Subject"; addheader "Subject" "Testing"; """
Putting the "deleteheader" above the "if body" causes it to be deleted correctly. Although I've used Subject in this example it happens for other headers.
Cheers, Dan
Op 5/1/2013 3:18 PM, Daniel Piddock schreef:
Hi,
We're using Dovecot 2.1.16 with Pigeonhole 0.3.4. Mail is delivered by exim4 to the dovecot LDA.
Sometimes users are unable to delete headers. I've narrowed the problem down to preceding the deleteheader command with a query against the body.
A simple example: """ require ["body", "editheader", "fileinto"]; if body :contains "!TEST!" { fileinto "INBOX.Trash"; stop; } deleteheader "Subject"; addheader "Subject" "Testing"; """
Putting the "deleteheader" above the "if body" causes it to be deleted correctly. Although I've used Subject in this example it happens for other headers.
Are you sure the fileinto; stop; part is not being executed? In that case this behavior whould be correct.
Regards,
Stephan.
On 01/05/13 14:51, Stephan Bosch wrote:
Op 5/1/2013 3:18 PM, Daniel Piddock schreef:
Hi,
We're using Dovecot 2.1.16 with Pigeonhole 0.3.4. Mail is delivered by exim4 to the dovecot LDA.
Sometimes users are unable to delete headers. I've narrowed the problem down to preceding the deleteheader command with a query against the body.
A simple example: """ require ["body", "editheader", "fileinto"]; if body :contains "!TEST!" { fileinto "INBOX.Trash"; stop; } deleteheader "Subject"; addheader "Subject" "Testing"; """
Putting the "deleteheader" above the "if body" causes it to be deleted correctly. Although I've used Subject in this example it happens for other headers.
Are you sure the fileinto; stop; part is not being executed? In that case this behavior whould be correct.
Hi Stephen,
I'm totally sure. The mails end up with two Subject headers so the addheader line is executing correctly. The bug doesn't even need the fileinto and stop lines, I'm not sure why I included them.
Cheers, Dan
On 5/1/2013 4:58 PM, Daniel Piddock wrote:
On 01/05/13 14:51, Stephan Bosch wrote:
Hi Stephan,
I'm totally sure. The mails end up with two Subject headers so the addheader line is executing correctly. The bug doesn't even need the fileinto and stop lines, I'm not sure why I included them.
Hmm, bizarre. I cannot reproduce the problem at this end. Could you send me the following:
- dovecot -n output
- an example message that triggers this behavior
Could you try to reproduce this with the sieve-test tool?
Regards,
Stephan.
On 5/1/2013 9:23 PM, Stephan Bosch wrote:
On 5/1/2013 4:58 PM, Daniel Piddock wrote:
On 01/05/13 14:51, Stephan Bosch wrote:
Hi Stephan,
I'm totally sure. The mails end up with two Subject headers so the addheader line is executing correctly. The bug doesn't even need the fileinto and stop lines, I'm not sure why I included them.
Hmm, bizarre. I cannot reproduce the problem at this end. Could you send me the following:
- dovecot -n output
- an example message that triggers this behavior
Could you try to reproduce this with the sieve-test tool?
Ok, this should fix it:
http://hg.rename-it.nl/dovecot-2.1-pigeonhole/rev/0163c45094a3
Regards,
Stephan.
On 02/05/13 20:48, Stephan Bosch wrote:
On 5/1/2013 9:23 PM, Stephan Bosch wrote:
On 5/1/2013 4:58 PM, Daniel Piddock wrote:
On 01/05/13 14:51, Stephan Bosch wrote:
Hi Stephan,
I'm totally sure. The mails end up with two Subject headers so the addheader line is executing correctly. The bug doesn't even need the fileinto and stop lines, I'm not sure why I included them.
Hmm, bizarre. I cannot reproduce the problem at this end. Could you send me the following:
- dovecot -n output
- an example message that triggers this behavior
Could you try to reproduce this with the sieve-test tool?
Ok, this should fix it:
http://hg.rename-it.nl/dovecot-2.1-pigeonhole/rev/0163c45094a3
Regards,
Stephan.
That appears to have fixed the issue in my testing.
Thank you for the quick patch.
Dan
participants (2)
-
Daniel Piddock
-
Stephan Bosch