Mauro,
OK for me...
Happy holidays and happy christmass
Thanks
Vadim
Mauro Sérgio Ferreira Brasil wrote:
Hi Vadim,
Sorry by the late answer, but I was extremelly occupied here with other
demands.
Regarding this bug fix, considerations follow below:
1) Ok;
2) I would preffer to put "ph_clear_msession_streams_fmtps" method
right before the copy block so I could use the same method that makes
the final deallocation of "fmtp" fields on "ph_msession_s" clean up.
Any problems ?
3) This method already exists and will be used on item 2 as well;
I've made some additional changes on "ph_update_media_payloads_fmtp"
and "ph_ilbc_get_frame_size" to make a more accurate payload id
validation of "fmtp" field to avoid wrong mode matching.
Regarding the patch to allow the negotiation and use of different
packetization on incoming and outgoing channels, I'll take a little
longer to post news and finally delivery it, because I'll be on
vacation (starting tomorrow.. uhuuuu!!!!), and I'll be back only at day
5 of January.
Thanks and best regards,
Mauro.
Vadim Lebedev escreveu:
Hi Mauro,
You're right this is a bug.
I suggest following approach.
1) Replace the offending assignement by a call which copies all
fields in the structure
field by field, execpt the fmtp field.
2) ph_update_media_payloads_fmtp should free the fmtp fields to which
it assigns value
3) upon call termination the fmtp filels should be freed too
Thanks
Vadim
Mauro Sérgio Ferreira Brasil wrote:
Hello
there!
Vadim,
Thanks for the answer on another email.
I'll have a good look on it as soon as I get rid of this problem.
We had some problems here with patch to handle "mode" parameter of
"fmtp" attribute for iLBC payloads.
The problem is on lines below from "ph_call_media_start" method:
msp->opayloads[0].number = ca->audio_payload;
ph_parse_payload_mime(&msp->opayloads[0],
ca->audio_payload_name, 8000, 1);
msp->ipayloads[0] = msp->opayloads[0];
ph_update_media_payloads_fmtp(je, msp);
During some additional tests, I've noticed that method
"ph_call_media_start" is called more than once per session. For this
reason, I've changed method "ph_update_media_payloads_fmtp" to validate
the "fmtp" field of "ph_media_payload_s" structure before reassign it.
Unfortunately, I still have a problem with the assignment made on first
3 lines.
It leads to a overriding of "msp->ipayloads[0].fmtp" value, that is
a "char *", with "msp->opayloads[0].fmtp".
Is there some possibilities to avoid this situation:
1- Use the same validations present on
"ph_msession_audio_stream_start" method that indicates the stream will
be reused to avoid the above assignment;
2- Use "ph_clear_msession_streams_fmtps" to clear the values of "fmtp"
fields from "ph_media_payloads_s" before the assignment and new update
of their values;
What do you think is the better choice ?
Please feel free to send any suggestions.
Thanks and best regards,
--
TQI - Technology and Quality on Information
| At.,
|
 |
| Technology
and Quality
on Information |
| Mauro Sérgio
Ferreira Brasil |
| Coordenador
de Projetos e
Analista de Sistemas |
| + [email protected] |
| : www.tqi.com.br |
| ( + 55 (34)3291-1700 |
| ( + 55 (34)9971-2572 |
--
TQI - Technology and Quality on Information
| At.,
|
 |
| Technology
and Quality
on Information |
| Mauro Sérgio
Ferreira Brasil |
| Coordenador
de Projetos e
Analista de Sistemas |
| + [email protected] |
| : www.tqi.com.br |
| ( + 55 (34)3291-1700 |
| ( + 55 (34)9971-2572 |
|
_______________________________________________
QuteCom-dev mailing list
[email protected]
http://lists.qutecom.org/mailman/listinfo/qutecom-dev
