June 10, 2008
View 78a65d14d4

- publishing up_since as core variable

Tomas Mandys authored on 10/06/2008 15:03:35
View 195e57116d

more verbose comment

Libor Chocholaty authored on 10/06/2008 12:30:38
June 9, 2008
View 5fd534209b

Man page for acc_db. Some corrections to stylesheet and the page for auth_db.

Martin Hoffmann authored on 09/06/2008 15:33:52
View 92e5ce8e9c

Stylesheet for docbook-based man pages. Man page for auth_db.

Martin Hoffmann authored on 09/06/2008 09:18:33
June 6, 2008
View b0c5ec1d3e

- parse_nameaddr doesn't log any error messages anymore (sometimes it's called to check if a string is in nameaddr format, for example from the avp module) - if parse_nameaddr fails in parse_rr, parse_rr will log the failed "uri".

Andrei Pelinescu-Onciul authored on 06/06/2008 12:12:33
View 7bba6b0eb1

- be more verbose when parse_nameaddr fails (show the guilty string). Patch from Alfred E. Heggestad <aeh@db.org>.

Andrei Pelinescu-Onciul authored on 06/06/2008 11:49:18
View 92060a0ea0

- standalone configuration file parser, this is an extended version of the parser from the tls module. This version can be used from other modules too, currently used by tls and ldap modules, in the future to be used by bdb, mysql, postgres, flatstore, and others.

Jan Janak authored on 06/06/2008 00:00:21
June 5, 2008
View a9ef106066

- don't path libdl and libresolve to modules by default (thanks to Marcus Better <marcus@better.se>)

Andrei Pelinescu-Onciul authored on 05/06/2008 16:36:58
June 4, 2008
View c1deee7e90

- T_CANCELED is set for any E2E CANCEL, making t_is_canceled() usable for more then checking for out fo order CANCELs (experimental)

Andrei Pelinescu-Onciul authored on 04/06/2008 13:13:54
May 31, 2008
View d853fdab95

- fixed indentation

Jan Janak authored on 31/05/2008 17:50:34
May 30, 2008
View 0dbf8ccae0

- ver. no ++

Andrei Pelinescu-Onciul authored on 30/05/2008 21:11:53
View c1d3d33509

- fix: in some corner cases where send blocks for a long time or when final reply or on send callbacks take too much time to execute for a transaction for which the final reply timer just expired, it is possible that the wait handler would execute _before_ we are finished with t (e.g. before all the callbacks are called). Because presence on a timer list is not ref. counted and we optimize deletion by allowing unlink_timer() not to wait for the fr_timer to finish (timer_allow_del()) this would mean the transaction will be deleted while still in use. The fix makes sure the wait timer is always started after we're not looking at t anymore (an alternative would be to remove timer_allow_del() from retr_buf_hanlder(), but this could cause fast timer slowdowns).

Andrei Pelinescu-Onciul authored on 30/05/2008 21:10:53
View f492982fea

remove trailing space

Alfred E. Heggestad authored on 30/05/2008 07:50:06
View 1b5741d0b9

fix bug in ser-oob introduced by beautification

Alfred E. Heggestad authored on 30/05/2008 07:40:33
May 29, 2008
View d7cdf2504a

Beautification of ser-oob.cfg.

Martin Hoffmann authored on 29/05/2008 16:21:02