Older blog entries for walken (starting at number 18)

Got my new machine, an athlon 950 (with asus a7v motherboard). Works nicefully :) I got a few minor problems with it though, installing a debian (woody) system on it :

- The two ATA/100 channels were not recognized by any of the debian rescue discs. So linux would not see my hard drive at all... for now I just plugged it on one of the ATA/66 channels, I'll try to see if linux can support the ATA/100 thing when I have everything else working. The controller is a promise PDC20265.

- I cant get XFree4 to work yet :( I dont know if this is because I have a matrox G450, or if it is because of debian. The best results I got till now is to get a garbled screen (the synchro frequencies are apparently very wrong) and have the X server quit, while leaving the screen garbled. Ctrl-alt-del from the keyboard still works. Huh :-/

If anyone else has an a7v or a g450, I'd love to know how to make it work :)

On the mpeg2dec side, I implemented support for field pictures, so now I want to do some video output lib changes and then releases 0.2.0. The libvo changes are a major subject of dispute on the livid lists, which is just not fun for anyone involved :-/

15 Nov 2000 (updated 15 Nov 2000 at 21:47 UTC) »

Sam (sh), I just love your new hammer design.

Finally made a stable mpeg2dec 0.1.7 release. Party time :)

That made me learn about autotools stuff. I didnt think it'd be so long to just take the current, working stuff and release it. In practice its not so easy, you want to update all file headers to put the GPL license stuff, have all the right README and INFO files, make sure it compiles everywhere, test, blah blah blah... Well I think future releases will be much easier to put out now that I've done that once.

Now that this is done I'll have time to work on features again :) beginning with support for field pictures...

Not been that much active this month.

libmpeg2 has been progressing slowly. Right now we have an mpeg2 conformance test suite and we seem to be conformant on mpeg1 streams, and mpeg2 when they don't use DMV or field pictures. Which means we are doing OK on most DVD streams. I guess I should bump the version number and make a freshmeat release sometime. The work on libvo and/or synchro has not progressed much though, which is quite sad...

I bought an used car and spent much of the week driving all around the SF bay to see friends. Fun :)

Damn ! I havent posted here for one month.

Lots of stuff have happened in the libmpeg2 since last time. I checked in the work I did on the slice decoder, basically a big restructuration of the code. Then I corrected lots of small details to make them actually comply with the standard. Then I rewrote the parsing code and we gained about 20% in decoding speed. And then MPEG1 support went in. Yummy :-)

I declared myself a lead developper. Advogato is good for your ego :-D

The next big milestones for the LiViD project will be to switch to a new API between the codecs and the output layers, and then to work on the synchronisation mechanisms. These times it seems that more people are coming to the project and trying to help. That's good. I'm trying to send detailed proposals for the output layers work, but this kind of coordination work always takes more time that you would want. It seems to be worth it though.

On the libmpeg2 side of the project, the next steps will be to implement field pictures and then the missing MC modes. The output layer work has a higher priority though, so I guess in the next days I'll not be progressing very fast on libmpeg2 proper.

Finally found a house to rent in oakland. Yay :) also opened my bank account, blablabla. Next thing is to buy a car, which also requires getting a californian license and an insurance. I'm sure I will love all this paperwork.

The NT on my new computer is half-dead already. It almost works, but for some reason it behaves like if I had some control key stuck, or I pressed the wrong mouse buttons, or I dont know what - I dont use NT much, so I dont know how you could make it not work like that. It just behaves in a very stupid way - when I click on an icon it gives me a popup menu instead of selecting it, etc... very unuseable. I have no idea when and why it started to act this way. Oh well, its only at work, so I will just have to call someone and watch him getting mad. I guess that could be fun.

It worked friday and monday, heh, two days, its not so bad for NT :-)

On the mpeg2dec side : the IDCT scaling thing is checked in, the slice rewrite is done, submitted, but not checked in yet, and I think we've made great progress since the OLS. Soon we should be able to try the mpeg compliance test bitstreams. Yummy ! :) I have not progressed any bit on the synchro stuff though.

Trying to find a cool room (with roomate) in the SF / oakland / berkeley / alameda area. Hasnt had much success yet. I have to leave my temporary appartment this weekend, so I guess I will have to spend a few days at the hotel. Oh well.

My rewrite of the mpeg2dec slice.c code seems to get into shape. I'm very happy at the look it has now - much more straightforward than any other mpeg code I have seen. I'm proud :) I will check this in sometime soon, perhaps this weekend if I'm not moving at that time.

Used my CVS write access for the first time. Yay :) That was not too big of a patch, basically a rewrite of the headers parsing functions using hand-parsing instead of the bitstream functions, and a removal of all the stuff we were parsing but never using for decoding. I think it makes things a little simpler to follow. I will have to do some changes in slice.c too, but I have carefully avoided that for now :)

Stuff to work on :

* get rid of the infamous IDCT scaling

* rewrite a lot of the slice.c code, right now we first parse everything and then run the MC/IDCT code, and a lot of the tests are duplicated between the parse side and the MC/IDCT side. I think the code will be easier to follow after I change this.

* synchro stuff, but I think there will be a lot of bitching between aaron and me on this one. It will not be a huge chunk of code but it is not easy to get right.

Back from OLS. That was great ! The confs were interesting, but of course the main discussions were between them and at the parties every evening... :)

Also I finally met aaron, which was probably half of the reason why I went there in the first place. I may get CVS access to the mpeg2dec tree soon, which should help a lot. We still dont quite understand each others views on the synchro stuff though, so I guess we will have to try things and compare code instead of ideas. We will get there eventually.

I still dont have my PIII at work, and development on this ppro 200 is a bit painfull. I still get a full 6 fps decoding rate though :->

Heading to OLS. I wish I would be there already. Oh wait, there would be nothing to do yet. I wish it would be tomorrow then :)

My slice.c /getvlc.c rewrite is progressing, but slowly. I'm trying to shorten the common pathes, with some success I think. It does break a few abstractions (like putting huffman parsing in separate routines), but it is still kind of clean. This portion of the mpeg standard is quite convoluted though, and it is not always easy to factor things out in a nice way. Lots of work to do in there, we'll see what happens :)

I also fixed a few non-conformance points (which added back some of the complexity, grumph) and potential buffer overflows...

I still dont have my P3 computer, so I'm using a ppro 200 for coding. I get about 6 fps when I'm lucky. yay :)

9 older entries...

New Advogato Features

New HTML Parser: The long-awaited libxml2 based HTML parser code is live. It needs further work but already handles most markup better than the original parser.

Keep up with the latest Advogato features by reading the Advogato status blog.

If you're a C programmer with some spare time, take a look at the mod_virgule project page and help us with one of the tasks on the ToDo list!