[asterisk-dev] State of parking code and bug 17680

[Dan Austin]

Wedhorn wrote:
> On Tue, 27 Jul 2010 12:33:20 -0700, Dan Austin <Dan_Austin at Phoenix.com> wrote:
>> My last question is sadly along the lines of C101, but I am running out of
>> reasonable google queries to teach myself what I need to fix bug 12324-
>> 

> Bug 12324 is a horrible little beast. While technically the issue is fairly
> simple, the structure of chan_skinny makes fixing it difficult, basically 
> because there is no consistent subchannel referencing in chan_skinny. For
> info, bug 12324 was fixed by the defunct bug 14997, and should be fixed when
> we fix up the sub handling stuff (work started but then stalled on bug 16980).


I played with it a bit today and the bitmask code was not going to do what I wanted.
I ended up adding a 16 element array of uint_32t to the device structure and storing
subchannel reference numbers.

This had a couple of benefits, I could send unique call 'counts' to the phone
display-
	1.  Play three calls, putting each on hold
	2.  Hangup the 2nd
	3.  The display shows calls 1 and 3
	4.  Place a new call and it gets put in slot 2
(current code has all calls as 00 or 01, I forget)

The display count is the 1st uint32_t labeled space in the callinfo structure

Events from the phone do not include the call 'count', but the call reference,
which I then use to clear the slot in the array.

I don't like the extra memory requirement, but I do not see another way to
relate a line instance and call reference to a specific call number on the phone.
The events coming from the device do not have the call number.

> Personally I'm of the opinion that 12324 be left on hold until we clean up the
> sub handling and referencing, at which point the fix becomes trivial.

> Damien Wedhorn

> PS If you have some time and can test 16980 (hope it still patches against trunk),
> we can progress the sub handling fixup.

I'll try to make some time this week to poke at it.

Dan