Summary: | ASTERISK-13063: "empty" overlap-dial-in doesn't work | ||
Reporter: | Matthias Urlichs (smurfix) | Labels: | |
Date Opened: | 2008-11-12 14:41:30.000-0600 | Date Closed: | 2008-11-13 13:17:46.000-0600 |
Priority: | Major | Regression? | No |
Status: | Closed/Complete | Components: | Core/PBX |
Versions: | Frequency of Occurrence | ||
Related Issues: | |||
Environment: | Attachments: | ( 0) bug13891.patch | |
Description: | Some PRI callers (like the ancient telephone system we want to replace) always send an empty SETUP message, and provide the actual number later. Unfortunately, pbx_find_extension() is set to return NULL if it sees an empty extension, thus preventing the call from ever completing. I noticed the problem with DAHDI, but I suspect it's not limited to that. ****** ADDITIONAL INFORMATION ****** The change was introduced in SVN change 116297. I don't know the rationale for that, but whatever problem that part of the patch fixed needs to get fixed in a different way. Empty extensions are perfectly legitimate in overlap-dial-in situations. | ||
Comments: | By: Leif Madsen (lmadsen) 2008-11-12 15:03:15.000-0600 I'm assigning this issue to jpeeler because svn log seems to tell me that's who made the modification at svn rev. 116297 By: Jeff Peeler (jpeeler) 2008-11-12 18:17:19.000-0600 Candidate patch uploaded. I feel like this might be what is causing the problem since the other changes were specifically for parking. I remember very briefly talking to murf about it as it did prevent a crash in some scenarios. By: Matthias Urlichs (smurfix) 2008-11-12 23:38:27.000-0600 That's exactly the change I applied to get the whole thing working (which should surprise nobody ;-). It remains to be seen whether the original problem is affected by this; I assume that there must have been a reason for that change ... By: Jeff Peeler (jpeeler) 2008-11-13 13:09:57.000-0600 The original problem was "when # is pressed and nothing is pressed, the code is currently trying to match on the extension '' and is falling through". (Extracted from my chat logs - May 13. I should have done a better commit message.) Reverting the code does not seem to have any negative effects so I'm going to take it out. I originally thought the change was OK to make as I thought it was mirroring the 1.4 behavior of pbx_find_extension returning null when an extension string of zero length is searched. More importantly perhaps I was not aware at the time that an empty extension was ever possible. By: Digium Subversion (svnbot) 2008-11-13 13:17:45.000-0600 Repository: asterisk Revision: 156649 U trunk/main/pbx.c ------------------------------------------------------------------------ r156649 | jpeeler | 2008-11-13 13:17:45 -0600 (Thu, 13 Nov 2008) | 6 lines (closes issue ASTERISK-13063) Reported by: smurfix This reverts a change I made in 116297. At the time it seemed the change was required to solve an issue with attempting a transfer but then letting it timeout without dialing any digits. However, I didn't realize that having an empty extension was possible. I'm removing the immediate return that was added in pbx_find_extension if the extension is null. ------------------------------------------------------------------------ http://svn.digium.com/view/asterisk?view=rev&revision=156649 |