[asterisk-dev] Muting a noisy warning in a specific case
Tilghman Lesher
tilghman at mail.jeffandtilghman.com
Fri Sep 4 12:36:22 CDT 2009
On Friday 04 September 2009 10:05:10 Tilghman Lesher wrote:
> On Friday 04 September 2009 05:43:20 Pavel Troller wrote:
> > Hi!
> > I'm using asterisk 1.6 (now it's 1.6.1.6) and my log was permanently
> > filled with the following warning:
> >
> > [Sep 4 10:28:20] WARNING[19721] pbx.c: The application delimiter is now
> > the comma, not the pipe. Did you forget to convert your dialplan?
> > (Set(CDCT=(^[a1-8].*$|^9[0-2].*$)))
> >
> > As you can see, this warning is issued improperly, because the "pipe
> > symbol" is a part of a regular expression, which is assigned to a
> > variable, and not a part of a general command syntax. AFAIK, the Set()
> > application syntax is very clean and doesn't allow neither commas nor
> > pipes to regularly appear in it. It means, that any pipe/comma present in
> > Set() is a part of the value being assigned, and this warning is wrong.
> > For every call, I get 3 such warnings (my extension macros are
> > relatively complex). I've patched pbx.c by adding ' &&
> > strcasecmp(app->name, "set")' to the test expression (there is a test for
> > the System() app already in place) somewhere around line 957 in pbx.c.
> > Would it be acceptable to put this test regularly to the sources ? Maybe
> > not just me would appreciate it.
>
> Boy, I wish that were true. Unfortunately (and I consider this a
> misfeature), Set in 1.4 and previous allowed setting multiple keys, by
> separating the key/value pairs with a pipe. The reason I consider this a
> misfeature is that it prevented the user from putting a pipe within a
> variable value.
After thinking about this a bit more, I've added the ability to turn off that
warning with the dontwarn option in asterisk.conf. A note to that effect has
been added to UPGRADE-1.6.txt.
--
Tilghman & Teryl
with Peter, Cottontail, Midnight, Thumper, & Johnny (bunnies)
and Harry, BB, & George (dogs)
More information about the asterisk-dev
mailing list