[asterisk-bugs] [JIRA] (ASTERISK-29267) H.263+ Format Attribute Module: not RFC 4629

[Alexander Traud (JIRA)]

Alexander Traud created ASTERISK-29267:
------------------------------------------

             Summary: H.263+ Format Attribute Module: not RFC 4629
                 Key: ASTERISK-29267
                 URL: https://issues.asterisk.org/jira/browse/ASTERISK-29267
             Project: Asterisk
          Issue Type: Bug
      Security Level: None
          Components: Resources/res_format_attr_h263
    Affects Versions: 18.2.0, 16.16.0, 13.38.1
            Reporter: Alexander Traud


Asterisk 11 introduced a format attribute module for H.263+. The idea is to pass the format parameters (line “fmtp” in SDP) from the originating call leg (caller) to the called leg (callee). The module is based on [RFC 4629|https://tools.ietf.org/html/rfc4629#section-8.1.1]. It parses all mentioned parameters and then generates one line “fmtp” again.

However, there are bugs and limitations:
# CPCF is simply missing
   OK, nobody (?) is using that; forget about it.
# CIFs are printed with MPI=0 when not specified
   That is a violation because MPI can be 1 to 32 only.
# K and N are printed with =0 when not specified
   That is a violation because 0 is not a defined value.
# BPP is printed with =0 when not specified
   although 0 seems to have a meaning.
# MaxBR is simply missing
   OK, not specified RFC 4629, but some implementations use it.

2-5 can be solved by applying the same unset/nonzero approach as in the format attribute module for H.264. See the attached patch.

However, however, RFC 4629 states: “Parameters offered first are the most preferred picture mode to be received.” This can be worked-around for the normal CIFs if one assumes that larger is better. In that case, the order of the CIFs in {{generate_sdp_fmtp(.)}} is simply reversed. See the attached patch.

However, however, however, RFC 4629 includes that CPCF and allows CUSTOM. Those can be better or worse than other picture modes. This cannot be incorporated that easily.

*Long story short*
Saving the whole fmtp as one line, coping that over, rather than parsing and generating all parameters; wouldn’t that make much more sense?



--
This message was sent by Atlassian JIRA
(v6.2#6252)