[asterisk-dev] [Code Review] Add #includeif statement

[Paul Belanger]

On 11-10-05 10:50 AM, Tilghman Lesher wrote:
> On Wed, Oct 5, 2011 at 9:47 AM, Paul Belanger<pabelanger at digium.com>  wrote:
>> On 11-10-05 10:41 AM, Tilghman Lesher wrote:
>>>
>>> On Wed, Oct 5, 2011 at 9:32 AM, Paul Belanger<pabelanger at digium.com>
>>>   wrote:
>>>>
>>>> On 11-10-05 10:21 AM, Jason Parker wrote:
>>>>>
>>>>> On 10/04/2011 07:47 PM, Matt Riddell wrote:
>>>>>>
>>>>>> What was the original reason that #include started blocking the loading
>>>>>> of
>>>>>> modules when the included file was missing?  I've always hated that.
>>>>>>
>>>>>
>>>>> Take the following config setup as an example.
>>>>>
>>>>> sip.conf:
>>>>> #include mypeer.conf
>>>>> #include yourpeer.conf
>>>>>
>>>>> mypeer.conf:
>>>>> [mypeer]
>>>>> type=peer
>>>>> host=10.0.0.2
>>>>> #include mypeer-secret.conf
>>>>>
>>>>> yourpeer.conf:
>>>>> [yourpeer]
>>>>> type=peer
>>>>> host=10.0.0.3
>>>>> #include yourpeer-secret.conf
>>>>>
>>>>> myypeer-secret.conf:
>>>>> secret=applesauce
>>>>>
>>>>> yourpeer-secret.conf:
>>>>> secret=embiggen
>>>>>
>>>>>
>>>>> The astute may have noticed that there is a typo in the
>>>>> mypeer-secret.conf.
>>>>> What would happen if we allowed the config parser to continue?  Very bad
>>>>> things.
>>>>>
>>>>> This is just one of several scenarios that I could think of.  It makes a
>>>>> lot of
>>>>> sense to me that if somebody uses a #include, then they absolutely must
>>>>> have
>>>>> that file - even if it's empty.
>>>>>
>>>> Right, and this approach for #include is fine to be honest.  I don't
>>>> think I
>>>> would suggest changing that functionality.  However, if you rely a lot on
>>>> include files for your asterisk setup, something we currently do with the
>>>> test-suite, having a lot of empty files lying around clutters up the file
>>>> system.
>>>>
>>>> One of the main reasons #includeif / #sinclude is being suggested is to
>>>> help
>>>> an administrator, who understands the differences between the two, cut
>>>> down
>>>> on the amount of files that needs to be managed (even if they are empty).
>>>
>>> If that's your intent, why not just use the file globbing already in
>>> the source, i.e.
>>> #include sippeers-*.conf
>>>
>> I can try again, but I was not able to get this working.  The module would
>> still fail to load.
>
> Clearly, if something is broken, instead of fixing it, we should
> instead code a whole new feature that accomplishes the same thing.
> Right?  :-P
>
Fix it?  Pfft :)  To be fair, I didn't know it worked like that. 
Looking into the code now to see what is wrong.

-- 
Paul Belanger
Digium, Inc. | Software Developer
twitter: pabelanger | IRC: pabelanger (Freenode)
Check us out at: http://digium.com & http://asterisk.org