Slashdot Mirror
MPlayer 0.90 released; MPlayer Maintainer Leaves
Viqsi writes "459 days after the previous stable release, the MPlayer folks have finally released stable version 0.90. With this done, A'rpi (th head maintainer) is leaving the project, citing too-much-free-time-forever-lost issues, and the team is looking closely at revising the way the project is managed as a result. Here's hoping some improvements come out of this."
.90 has been a long time in coming, but the wait was well worth it. I continue to be astounded by what mplayer and mencoder are capable of, and I shudder to think of what my Linux movie watching experience would be like without them. I hate to sound like a cheer leader, but I just don't think enough can be said about the fine work that A'rpi and Co. have produced over the years. In addition to our beloved kernel, it's always nice to have examples of open source software that so readily stomps into irrelevance its closed source competition. Good luck to A'rpi in whatever the future holds, and a thousand thanks for your contributions to the community.
Not many know this but texstar has a awesome package for embedding Mplayer into Konqueror. This is just awesome.
Big thanks to the developers., for this one.
Of course, I don't mean to complain, for goodness sakes. A'rpi has done an amazing job. I recently introduced a friend of mine to linux, and he was awe-struck by the huge functionality and flexibility of mplayer. I will always be greatfull for the development that has gone into this wonderful program that I use every day.
Mplayer only just recently got inverse-telecine, which is a Good Thing for those of us who like to archive a lot of shows.... I used to have to use virtaldub under wine if I wanted to get IT done right [for truly treasured movies... the rest I just deinterlaced and accepted the quality loss].
The only remaining quibble I have with mplayer, actually most specifically with mencoder, is the lack of any threaded/forked/etc rendering pipeline. I am somewhat in the minority in having a SMP system, but there are a good deal of us out there. It causes me such pain to not have quite enough cpu to do certain realtime effects while encoding, to fall behind, while I see CPU0 pegged to 100%, and CPU1 just sitting there idle.
There was a big occasion for disagreement a while back, when someone tried to get some threading into mplayer, even had working patches. A'rpi refused. He had somewhat of a point; the context switching overhead actually wastes cycles on a single processor. [As well as flushing the cache at inoportune times]. Threading on a uniprocessor system would only really help with I/O latencies, but mplayer has great cacheing, and manages well, even with just a single context.
I'm torn between being hopefull that maybe there will be more openess to future improvements such as SMP support, but I'm also sad to see such a wonderfull developer throw in the towel. MPlayer probably wouldn't have happened without him.
---
the pen is mightier then the sword. the sword is mightier then the court. the court is mightier then the pen.
And I really hope they loose theit fear of Version number 1.0. I am pretty amnazed that Arpi leaves without having a half decent 1.0 out. gui is still disabled by default. Why? Largefiles: disabled. Why? The output mplayer sends to stdout is still a incredible mess. I wonder why he leaves the project without cleaning up these few last bits. In principle MPlayer is the most usable, stable, and featurefull media player for linux. Only thing is, it's a mess.
I hope the first thing they do is clean up the code. MPlayer lost _many_ developers to xine lately. xine has not caught up to MPlayer in speed and number of supported formats. Also it seems to be still less stable and more vulnerable to broken video files, but the code base is MUCH more clean (xine sources can actually be called beautifull) Maybe this disruption of MPlayer development can also be seen as a chance for a more unified default mediaplayer for linux, i.e. xine.
The one thin xine is lacking is an encoder a la mencoder. But there is some development with enix. The design looks as clean and easy as xine and I am pretty confident that enix can catch up to mencoder in a short time provided that some more developers are interested.
Cheers
KdenLive/PIAVE - non-linear video editing
I think the last thing the world needs is another media framework. Not that it's a bad idea per se, or that gstreamer is bad. But there are too many perfectly good frameworks already. And they all have one thing in common: while they promise ultimate flexibility and all kinds of ubercool features, in practice, only a few combinations of plugins work really well. And even those would have worked better if they had simply been integrated into a single app.
What we need is code that does heavy lifting. Code that can read and write all sorts of file formats and that can do so at blistering speeds. But that requires mindnumbing attention to detail and painstaking testing on a wide range of different files and machines. So that most people prefer to indulge in the fantasy that all of this difficult work will just somehow evaporate, just as long as there is a good framework: "it's the bureaucracy, stupid". Of course, that rarely, if ever, pans out. At best, the grandiose Framework is reduced an API for manipulating media clips (QuickTime). At worst, it forever remains a flaky research project to satisfy the will-to-power of a few geeks and true believers (http://www.sourceforge.net).
Now that last part was a bit flamebaity. But what I'm trying to say is that instead of spending time on developing a "media framework" that can do "every conceivable thing", is almost always better to spend your time considering which of those "conceivable things" actually makes sense: and implementing that.
"How do I..." "RTFM IDIOT!!!11"
You mean THAT A'rpi ?
I think that attitude is something that easily happens if you invest too much time and devotion into a project, especially if that project is really rather complex and draws the attention of a lot of newbies to Linux who want to watch vidz. If you're trying to focus on the development with hundreds of dependencies to keep in mind and every day you get the same "How do I.." FAQs on the mailing list, I guess you tend to go down the "It's in the docs, check the README file", "Please read the docs, it's all in there", "PLEASE read the docs, we didn't write them for fun" "READ THE GODDAM FUCKING DOCS!!" route rather quickly.
Maybe projects like this should have separate PR people who are not as directly involved with development but know enough about the project to hand out clues to newbies, keeping the coders free to focus on deeper things.