Playing a lot of "poke the patches in review so they happen" last couple days. Yay patch wrangling! Still getting used to Phabricator as a task tracker; I don't like the search but its otherwise not bad, and in some ways much more flexible than Bugzilla.
Got very confused when I started getting compiler errors in a line that looked fine. Turned out what I thought was a speck of dust on the screen was a stray ` character...
Beginning to grok the CocoaPods package manager for Obj-C libraries; reworking my iOS Ogg Theora player demo so it's packagable as a drop-in widget!
Still needs work but it builds!
Evan Prodromou likes this.
MediaWiki dev summit in San Francisco... we're trying to set some future directions for cool and productive stuff. :)
Playing with https://github.com/trifacta/vega/wiki/Headless-Mode for Yuri's extension https://www.mediawiki.org/wiki/Extension:Graph
Has Slashdot gone super downhill or are my memories of the late 90s/early 2000s FOSS scene just tinged with rosy nostalgia?
http://tech.slashdot.org/story/14/11/16/1619248/gtk-developers-call-for-help-to-finish-cross-platfor... has literally TWO productive comments -- mine, and a single vague reply to mine. The rest, including all the ones upvoted as "insightful", are trash-talk about GTK+, Gnome, Red Hat, or systemd.
Anyway, I got the Mac OS X port of the GTK+ OpenGL stuff mostly working, with patches in bugzilla for gtk+ and cairo. I had fun writing it at least. :)
Evan Prodromou likes this.Slashdot has gone downhill, yes, but so have the GTK+/GNOME projects. They HAVE been removing features and APIs without deprecation in minor versions, and are hostile to contributions/patches (case in point: why did you need to pull working patches from Bugzilla? why haven't they been merged already?).
The easiest thing to do is just say that Slahdot is a bunch of haters (it is), but that's ignoring the huge PR problem the GTK+/GNOME projects have—one they refuse to acknowledge or do anything about.
email@example.com likes this.
New IE updates in latest Windows 10 preview build. Still no Web Audio but they assure us it's coming! :D http://blogs.msdn.com/b/ie/archive/2014/11/11/living-on-the-edge-our-next-step-in-interoperability.aspx
I was a little disappointed that the "IE dev channel" from a few months ago never got updated, but updates seem to be coming at a reasonable pace in the Windows Technical Preview now, which makes me a happy webdev panda.
Fell down a rabbithole researching possible freeish archival video formats thanks to wikivideo-l mailing list.
Couple links of interest:
^ MXF container format is apparently A Thing out there in video/film production and archival. Most of the commonly supported codecs for it seem to be variants on MPEG-4 AVC in intraframe mode, though, which is not a happy patent story.
^ Some folks are trying to make a free-codecs-only profile of MXF called MOX and create plugins for Adobe Premiere etc.
^ Library of Congress has been doing JPEG 2000-in-MXF for archival, and the fed government may be standardizing on something in that vein.
^ DNG-in-MXF format that Adobe apparently has been trying to make an openish standardish, but no idea if adoption is good.
In the meantime, the most interoperable pretty-good-quality-but-not-uncompressed-appears-too-be-free format I can find is Motion-JPEG in QuickTime. :P Premiere, Final Cut Pro X, and Pitivi consume it happily and Premiere and Pitivi will produce it easily, but I don't know if it scales to modern quality needs (eg, if you want 10 or 12 bit channels, alpha, or full chroma sampling, I don't think JPEG does it). Of course this is assuming there's no patent issues on the QuickTime container format itself. :P MJPEG will happily live in other container formats, but then you run into interop issues again...
ffmpeg will produce .mov or .mkv files with JPEG 2000 encoding as well but it's hella slow compared to encoding Motion-JPEG. Support apparently recently landed for MXF with JPEG 2000... but it's not in the latest release yet.
Mike Linksvayer likes this.
Finally got round to skimming the Media Source Extensions spec; it's starting to make some kind of sense to me. :) https://dvcs.w3.org/hg/html-media/raw-file/tip/media-source/media-source.html
Unfortunately the current spec doesn't allow you to, say, send raw uncompressed video frames and audio samples into a <video> element -- it wants compressed data chunks in a known format, and the browser implements demuxing and decoding just as it does for conventional streams referenced via an HTTP URL.
Perhaps one day. :)
Claes Wallin (韋嘉誠) shared this.
Updates for VLC now landing which'll make it super-easy for us to build a free-codec-only version of MobileVLCKit for embedding in our iOS Wikipedia app to do Ogg & WebM media playback, without the dreaded MP4 "internet broadcasting" patent licensing.
Awesome! Big thanks to the VideoLAN folks for helping us with this!
Hoping to try out the new TestFlight beta system for iOS 8 for our Wikipedia app... of course it's tied into iTunes Connect now (unlike the classic TestFlight system where we uploaded ad-hoc or 'enterprise' builds directly) and so I have to wait for some mysterious "processing" to finish on my build before I can invite testers. Oh, Apple.
Looks like the 2009 Wikimania presentation videos on Commons are buggy (I think they were chopped from streams in a weird way) and don't play in Chrome... or correctly transcode to WebM.
Found an easy way to rewrite the file structure with ffmpeg while keeping the original audio/video encoding intact (-c copy); oggz-chop will also add back a skeleton metadata track which ffmpeg doesn't know how to create. Notes at https://commons.wikimedia.org/wiki/File_talk:200908281410-Andrew_Lih-Challenges_of_Covering_the_Wiki...
Now I just have to automate the download/fix/reupload process on the other 82 presentations from that year... sure as hell not gonna do all those files by hand. ;)
Evan Prodromou likes this.
Even one of the original authors of MediaWiki has trouble uploading media files to Wikimedia Commons... http://magnusmanske.de/wordpress/?p=236
/me again ponders the transcoding-on-upload problem for mp3/mp4/etc files... is there some server-side solution with a 'safe' license if we don't expose the original files to redownload, or shall I just start work on a client-side upload/transcode helper app?
Hey, I always wanted to learn the encoding interfaces for Vorbis, Theora, and VP8 anyway... MORE WEEKEND & EVENING PROJECTS!
Turns out liboggz's seeking doesn't seem to be optimized to use ogg skeleton data anyway. :) Managed to find the well-hidden 'libskeleton' -- https://git.xiph.org/?p=libskeleton.git;a=summary which actually parses this data and includes a function to give you the last-keyframe offset for a given time position. Heyyyyy sounds handy!
(Note it doesn't clone over https. Bug filed.)
(And it doesn't compile with clang. One-line fix patch submitted.)