2010-03-10 16:47:28 -05:00
|
|
|
== Interface inconsistency ==
|
|
|
|
|
- enum_id member of ftdm_event_t is inconsistent. Most of the time is just for OOB events, the only other
|
|
|
|
|
type of event as of now is FTDM_EVENT_DTMF and is not using the enum_id member. I think we can get rid
|
|
|
|
|
of the FTDM_EVENT_DTMF and create ftdm_dtmf_event_t type instead of reusing ftdm_event_t
|
|
|
|
|
then ftdm_event_t would be renamed to ftdm_oob_event_t and the enum_id renamed to type, then ftdm_span_next_event()
|
|
|
|
|
will only return OOB events
|
|
|
|
|
|
2011-01-17 12:44:52 -05:00
|
|
|
- Deprecate last_error members.
|
|
|
|
|
It requires a lot of discipline to set the last_error string for every failure.
|
|
|
|
|
It does not add much value to the user either, most of the errors are criptic and
|
|
|
|
|
cannot be shown to end users, we already provide extensive logging for problem
|
|
|
|
|
troubleshooting.
|
2010-03-12 13:27:24 -05:00
|
|
|
|
2011-01-21 10:51:34 -05:00
|
|
|
- Implement threaded IO.
|
|
|
|
|
Currently IO modules only work on-demand, where the user (ie, FreeSWITCH) drives the read/write
|
|
|
|
|
of media. If the user stops reading, some functions are not possible
|
|
|
|
|
(DTMF detection or Hangup tone detection). It would be useful to implement a FreeTDM mode
|
|
|
|
|
where the media is driven by a group of threads that are always reading (and possibly writing)
|
|
|
|
|
then when the user does ftdm_channel_read(), the media would be read from the buffers filled
|
|
|
|
|
by the media thread and not from the underlying IO device, this gives a chance to FreeTDM to
|
|
|
|
|
still perform hangup detection or other media services even if the application is not reading.
|
|
|
|
|
|
