[XviD-devel] dev-api-4 this weekend

[suxen_drol]

On Thu, 20 Feb 2003 15:33:34 +0100 Edouard Gomez <ed.gomez at free.fr> wrote:
>
> @pete:  why  commiting   your  test  bed,  what  have   they  more  than
> xvid_decraw/xvid_encraw/(future) xvid_stat have not  yet ? I remeber you
> sent me these files long ago, and  iirc they do the same job. Correct me
> if i'm wrong.

yep they perform the same thing. i like rawenc, because its very simple.
someday i would like see a single "xvid" command line tool. ie.

% xvid [command] [options...] input [output]

% xvid enc pgm%03i.pgm stream.m4v		; basic encoding
% xvid enc --psnr pgm%03i.pgm			; xvid_stat
% xvid dec stream.m4v pgm%03i.pgm		; decode
% xvid test					; xvid test/bench

On Thu, 20 Feb 2003 17:11:08 +0100 elcabesa <elcabesa at inwind.it> wrote:
> ok so another question
> how a coder can know which xvid library is using?
> will api number change?

the new api will use slightly different public symbols to the existing
api. i plan to call them: xvid_global, xvid_encoder, xvid_decoder.
OR, should we respect our encore,decore heritage?

the new api will have a well documented version system, to enable
forward/backwards compatibility. if you dredge up some mails from
october-thru-decemember, you will see our discussion about versioning.

i consider libxvidcore and the api version to be the same thing. when
the new api is stabilzed, i intend to call it "v1.0.0".

On Thu, 20 Feb 2003 23:23:29 +0100 Edouard Gomez <ed.gomez at free.fr> wrote:
> I'm wondering why we have a test function that takes the param1 as a
> "what tests do we perform" flag but there are no documented flags in
> xvid.h. Could the one who commited tests do proper flags documentation
> and move them to xvid.h as required ?

this code is from my "new-api" tree. ~two months back, we were
experiencing problems with dcts and quant's above>14, so i commited
this test code. the 

my intention was to have something like xvid_bench inside xvidcore,
which can be compiled with a ./configure flag. however i never decided
on a proper xvid_test_t struct, or a console output mechanism (for
example: printf-like callback?)

since the XVID_TEST_xxx stuff is api related, this can be fixed as part
of the api upgrade.

-- pete; life is like a box of ammo