Re: NMS and export layer, the state of the union

On Nov 20, 2007 9:42 AM, Andrew Church <[email protected]> wrote:
> >While 1.1.0 is progressing slowly but steadly, I'm also restarting to work
> >on HEAD, thanks to some extra time kindly provided by Magic & Miracles dept. ;)
>
> So THAT'S where it went!  The instant I went to update from CVS, the free
> time I thought I had all vanished into thin air... oh well. ;)

Well, is not THAT big ;)

There is a bunch of sparse (minor fixes), that's the reason why so
much files are touched.
The highlights are:
- framebuffer core, and so decoder/frame_threads/encoder, was
SERIOUSLY rewrote, now
I like much more the code, the API and the design (and no, not because
I'm wrote it :P).
Any comment about the current code will be MUCH appreciated, since I
know there is still a lot of room for improvements.
- events (as in socket commands and/or signals) are handled by a
separate thread again (see transcode.c). I'm not a sponsor of this
design, I just seen that this seems to solve our problems in a fairly
clean way (and as a bonus, separate completely the encoder from the
events/sockets), but I'm open to other solutions as well
- ImageMagick modules where rewriten, switchig to new, higher level
MagickWand API.
- NMS encoder modules now have (or should all have) a private,
unformal, flush() operation.
Next ModuleSystem revision will require that.
- PVM3 code compiles again against libtc. Probably still broken at
runtime. We need a MAJOR redesign here.
- I've just started to address avilib refactoring task. Ugh.
- import_ffmpeg now uses libavformat, I've tested a few files with now
regression.

I think that's all.

Bests,

-- 
Francesco Romani // Ikitt