[XviD-devel] Development status
Edouard Gomez
ed.gomez at free.fr
Sat Apr 3 00:19:07 CEST 2004
Hey hey,
Writing an email to clear development status, tell where we are, and
where we would like to go.
TOC:
1. 1.0 Shedule.
2. CVS branches
3. Upcoming features in 1.1 or even 1.0.x
4. Arch/tla archive
/////////////////////////////////////////////////////////////////////////
1. 1.0 Shedule.
I think we covered all bugs found with the RC3 release. I propose
another RC for this week end. We will then give a 1 week or 2 week
timeslice to allow users find remaining bugs.
After this 1/2 week delay, if no bugs are found, RC4 is simply renamed
final. Else depending on bugs severity, a new RC or final will be
released. So in my opinion, XviD's schedule obeys the "when it's ready"
moto.
Are you all ok with this ?
/////////////////////////////////////////////////////////////////////////
2. CVS branches
Here's a summary of what branch is supposed to be, its state, and ...
blablabla, no more excuses for sysKin about "i forgot the name branch
to commit this stuff to" ;-)
** release-1_0-branch
With xvid 1.0.0 nearing completion, i created a dedicated branch to
continue bugfixing commits. All 1.0.x revisions will be tagged on
this branch.
** dev-api-4
This never ending branch served its purpose, bring us 1.0 base
code. So it's now declared officially dead, and no one is supposed
to ever commit stuff there anymore.
** HEAD
This sticky branch will be used to develop 1.1 code base. It's
already different than release-1_0-branch as pete commited his
brightness control code. The rule is simple, release-1_0-branch
bugfixing will always does its path to HEAD, non intrusive
features of HEAD can go to release-1_0. The non intrusive nature
of a patch is to be discussed by the submitter and the one that
volunteers 1.0 maintainership.
** April 1st tree
... hmm forget about that one, however the code is still available
on my site for people wondering how it was done.
/////////////////////////////////////////////////////////////////////////
3. Upcoming features in 1.1 or even 1.0.x
The fun part for hackers, some features are simply already in the patch
queue, others are part of a wishlist.
Patch queue:
- Complete postprocessing code (brightness is done, what is the dering
status ?)
- PPC port merge (already commited to my arch/tla head following
branch, i'll just wait for main author feedback about probable merge
mess)
- Fast ME refactoring (syskin may tell the status)
- RD optimized ME for BFrames (still sysKin stuff)
WishList:
- Clean the plugin mess, i think it's just better to export the
Encoder struct. However, this requires some big cleanup in the
structures, to make sure, the interface will not use duplicated
structures between internal and public API layers.
- My RC 2pass rewrite was just aiming at fixing VFW code
implementation errors. I know the ideas on which the code is based,
are way too simple to be true at low bitrates where distortion and
high unpredictability of frame size make it really having hard time
controling the bitrate efficiently. I'm open to new/less flawed
ideas.
- [Open discussion put here what you would like]
/////////////////////////////////////////////////////////////////////////
4. Arch/tla archive
I know arch/tla has a growing popularity amongst GNU/Linux developers,
so maybe someone is already using arch/tla with my archive. So I have
to advertise here what branches do what:
So my archive is still:
http://ed.gomez.free.fr/arch-repositories/ed.gomez@free.fr--2004-1/
http://www.sourcecontrol.net mirrors my archive just in case free.org is
down (quite often)
** xvidcore--devapi4-ppc--1.0
Used to dev the PPC port with christoph naegeli.
This work is already merged in xvidcore--head--0.0, so don't use
this branch anymore.
** xvidcore--head--0.0
Used to follow CVS head development.
** xvidcore--stable--1.0
Used to follow release-1_0-branch development.
** xvidcore--devapi4--1.0
As its CVS cousin, officially no more used.
Don't forget, this tla archive is maintained by hand (mines) so it can
be unsynced with CVS (in both ways, some stuff may be commited there
before it goes to CVS, and vice versa). But it presents the advantage
to be an atomic way to get xvid sources and report bugs for known tree
states.
This email is already too long, i hope someone will read it entirely at
least.
--
Edouard Gomez
More information about the XviD-devel
mailing list