[asterisk-dev] asterisk tests for libpri

[Kevin P. Fleming]

On 03/29/2011 12:11 PM, Tzafrir Cohen wrote:

> I was also surprised to see that the configure script tests 6 times for
> features that were added in tag 1.4.12-beta1 and 7 times for features
> added in tag 1.4.11 .
>
> Maybe some of those could be removed to slightly reduce the run time of
> the configure script?

Well, maybe. The idea with autoconf is that tests are done for specific 
features, not for versions, and there's no expectation that the presence 
of feature B implies that feature A is present (unless feature B relies 
on feature A). It may be most commonly true that if B is present A is 
also present because they appeared in the same release of the code, but 
it's possible that versions with B but not A could exist (especially 
since this is open source and users are free to modify the code to 
disable features if they have a need to do so).

>> If this sort of use case really exists, the most effective way to handle
>> it is to link libpri into chan_dahdi/sig_pri statically, so that it is
>> not possible for them to end up mismatched.
>
> Which means: when upgrading the version of libpri in a repository, be
> sure to rebuild asterisk right afterwards, and explicitly requier a
> newer version of the binary version.

No, that is not true. Putting a newer version of libpri into a 
repository will not result in Asterisk binaries being dependent on that 
version unless the Asterisk code was already capable of using the 
features in that newer version. If that is the case, then yes, if 
Asterisk is rebuilt, the Asterisk binaries will require a libpri version 
with the same features available as the one they were built against.

-- 
Kevin P. Fleming
Digium, Inc. | Director of Software Technologies
Jabber: kfleming at digium.com | SIP: kpfleming at digium.com | Skype: kpfleming
445 Jan Davis Drive NW - Huntsville, AL 35806 - USA
Check us out at www.digium.com & www.asterisk.org