[XviD-devel] [BUG?] intra meaning

[Edouard Gomez]

--qDbXVdCdHGoSgWSk
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
Content-Transfer-Encoding: quoted-printable

It seems one of my patches is breaking vfw because i used a free slot of
the intra value of pFrame.

When  i  searched  a  way   to  report  SVOPS  i  searched  through  the
encoder_encode_bframes  to see  what  intra  value was  free  to use,  i
noticed the value 3 was free so i mapped it to the SVOP type.

It seems  that assumption is  responsible of vfw breakage,  because this
value was used in vfw.

<quote from IRC - Koepi>
[01:22] <Koepi> "we defined that with sysKin and suxendrol some months
                ago" - if i was disconnected already and it didn't reach
		you

[Then some big big lag reported by my irssi program]

[01:26] <Koepi> ok, just because _you_ had nothing like that, you break
                usability for thousands of users? that's not fair
[01:26] <Koepi> nvops
[01:26] <Koepi> =3Dskipped frames
[01:26] <Koepi> fact is
[01:26] <Koepi> svop !=3D iframe, bframe, delay frame or not coded frame
[01:26] <Koepi> if you look at that listing, all frame types are
                exclusive
[01:27] <Koepi> your svop is an AND
</quote>

If i understand well you defined intras so that (type1 & type2) =3D=3D 0

Then i would like someone aware  of this rule to repair my damages which
are not  so big. Just search at  pFrame->intra in encoder_encode_bframes
and modify the else statements of such if conditions.

<code>
    if(pEnc->current->coding_type =3D=3D P_VOP)
        pFrame->intra =3D 0;
    else
        pFrame->intra =3D 3;
</code>

And   put   the  same   value   in   the  intra2coding_type()   function
(encoder.c:659)

And   please,  PLEASE   *use*  defines   instead  of   unbearable  magic
constants. Thanks a lot for the attention. Good night :-(

NB: next time i would like this kind of "voodoo mistery" explained in
    the code, else how the hell am i supposed to know something works
    like this and the other works like that ? Even christoph told me
    to use 3 for SVOPS because he was as aware as me of this exclusive
    rule.

--=20
Edouard Gomez

--qDbXVdCdHGoSgWSk
Content-Type: application/pgp-signature
Content-Disposition: inline

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.2.1 (GNU/Linux)

iD8DBQE+LJlGR5dTYz5sWMcRAtkUAJwIJtl5/FfFN4HO1ogCpve9TCksTQCeOqga
eaBHtBss133sgeRCaKIVYZE=
=62sp
-----END PGP SIGNATURE-----

--qDbXVdCdHGoSgWSk--