[MPlayer-dev-eng] Xine-lib Update Improves A Bit For Blu-ray

[Uoti Urpala]

On Tue, 2009-12-01 at 20:57 -0200, Dâniel Fraga wrote:
> 	It's a pitty when someone makes some fuss about a new feature
> and then after a while it gets cold... anyway there's one thing I do
> not understand about MPlayer: in the past, I don't know how long, but
> maybe 10 years ago, MPlayer was actively developed and there're lots of
> develoopers right? Or am I wrong? So what happened? Most of the
> developers just lost interest on MPlayer?

There was more development activity in the past, but it was not a
healthy project overall. The people who started MPlayer had enough skill
to hack things together, but they had almost zero clue how to manage a
project or _maintain_ a codebase, especially one of MPlayer's eventual
size. That resulted in a hacky codebase which made it increasingly hard
to do any new substantial improvements without breaking one of the
existing hacks. And new people joining the project without much
experience from other projects came to consider the bad development
practices as normal, making later improvement harder. So MPlayer went
from a project that definitely did not teach people any good practices
for working on a large project, to a project that not only required
working on a large codebase but a large _ugly_ codebase. At no point
have there been many developers actively and effectively working on the
kind of development that would fix existing issues and improve the
overall architecture and code quality.

It's actually not that easy to contribute to MPlayer. Most of the easy
and popular things have been implemented already. Many of the remaining
features would requite cleaning up some of the existing code first,
which requires more skill and understanding of the codebase. Look for
example at some of the recent subtitle-related patches. If MPlayer's
subtitle code was clean overall the authors of those patches would
probably have been able to integrate their features with little trouble.
But the original design and implementation of MPlayer's subtitle code
were complete crap, and although it has improved it's still not good. So
integrating the patches cleanly would require cleaning up some of the
existing code. Even if the authors were willing to do that themselves,
it requires more understanding of the sources to see what code _should_
be there instead of what _is_.

> 	Today, my impression is that MPlayer is not developed at the
> fast pace it was in the past. But maybe I am wrong...

Measuring by things like number of commits there's no doubt less
activity. OTOH the earlier kind of activity was not really sustainable.

I've also done development in the git repo that has had little
visibility on this list (and there's also a number of things by other
people).