GStreamer 0.10 Is Dead, But 1.0 Has Lots Of Good Stuff
Tim-Philipp Müller of Collabora delivered the keynote this morning on the second and final day of the 2012 GStreamer Conference. Similar to yesterday's keynote about GStreamer 1.0, Tim-Philipp Müller talked about how GStreamer 0.10 is dead and the future is with the soon-to-release GStreamer 1.0 and more of the future plans for this multimedia framework.
Here are some of my notes from this morning's keynote from San Diego:
- GStreamer 0.10 is dead. Müller acknowledged, "we haven't managed the transition very well [from 0.10 to 0.11/1.0." They may do bug-fixes if they can easily cherry pick it to 0.10, but there's no new development at all going into it. "Current 0.10 Git is a little unsavoury...It's not great." In terms of any future 0.10.x point releases, "we might do another 0.10 bug-fix release but no promises."
- Going from 0.10 to 1.0 is an evolution, not revolution. "Most things are still the same." For application developers the migration path to the new 0.11/1.0 API should be fairly easy while there is more pain exposed to GStreamer plug-in developers.
- New features of GStreamer 1.0 were highlighted like the dynamic pipelines, renegotiation support for all pipeline configurations, Table of Contents (ToC) support, VA-API hardware-accelerated video decoding, a video overlay composition API, better Windows plug-ins, better OS X plug-ins, the GStreamer SDK, etc.
- The GStreamer 1.0 status is that most all work has been completely ported, many elements from -bad also ported, many apps ported (namely the GStreamer GNOME applications) and are working, and porting apps was really easy going from 0.10.
- What's TODO for 1.0: the hard work has been done, conceptually the API is ready, and just some tinkering around the edges is finished. Almost everything has been ported.
- GStreamer 2.0 will be the next time there's an ABI break. There will be more bug-fixing going forward with 1.0.x point releases while GStreamer 1.1.x will be used for pre-releases. "We need to get bug-fixes out faster."
- Ongoing and future work for GStreamer includes in-vehicle infotainment (IVI), in-flight entertainment (IFE), DVB support / STBs, MPEG-TS support, and better connectivity (DLNA, SmartTV, etc). Some other future work also includes sub-titles using the overlay composition API, more application control, output-surface resolution dependent sub-title rendering, and support for more formats. Also being planned is better VDPAU video decoding support, better and more natural OpenGL integration, generic context sharing / distribution, and multi-view 3D video support. Some other items talked about were optimizations of new features, video parsing library improvements, GES improvements, better device discovery and probing, "smooth as butter" stream switching, play-list support, splitting the gst-plugins-bad module, Bluetooth plug-ins, and development tools/debugging improvements. For development tools/debugging there are hopes for better error reporting, tools collection, and ideas floating around about dot file dumps (dumping the GStreamer pipeline and also ideas for making it interactive).
Also be sure to read GStreamer 1.0 Is Looking To Finally Be Released Soon.
Here are some of my notes from this morning's keynote from San Diego:
- GStreamer 0.10 is dead. Müller acknowledged, "we haven't managed the transition very well [from 0.10 to 0.11/1.0." They may do bug-fixes if they can easily cherry pick it to 0.10, but there's no new development at all going into it. "Current 0.10 Git is a little unsavoury...It's not great." In terms of any future 0.10.x point releases, "we might do another 0.10 bug-fix release but no promises."
- Going from 0.10 to 1.0 is an evolution, not revolution. "Most things are still the same." For application developers the migration path to the new 0.11/1.0 API should be fairly easy while there is more pain exposed to GStreamer plug-in developers.
- New features of GStreamer 1.0 were highlighted like the dynamic pipelines, renegotiation support for all pipeline configurations, Table of Contents (ToC) support, VA-API hardware-accelerated video decoding, a video overlay composition API, better Windows plug-ins, better OS X plug-ins, the GStreamer SDK, etc.
- The GStreamer 1.0 status is that most all work has been completely ported, many elements from -bad also ported, many apps ported (namely the GStreamer GNOME applications) and are working, and porting apps was really easy going from 0.10.
- What's TODO for 1.0: the hard work has been done, conceptually the API is ready, and just some tinkering around the edges is finished. Almost everything has been ported.
- GStreamer 2.0 will be the next time there's an ABI break. There will be more bug-fixing going forward with 1.0.x point releases while GStreamer 1.1.x will be used for pre-releases. "We need to get bug-fixes out faster."
- Ongoing and future work for GStreamer includes in-vehicle infotainment (IVI), in-flight entertainment (IFE), DVB support / STBs, MPEG-TS support, and better connectivity (DLNA, SmartTV, etc). Some other future work also includes sub-titles using the overlay composition API, more application control, output-surface resolution dependent sub-title rendering, and support for more formats. Also being planned is better VDPAU video decoding support, better and more natural OpenGL integration, generic context sharing / distribution, and multi-view 3D video support. Some other items talked about were optimizations of new features, video parsing library improvements, GES improvements, better device discovery and probing, "smooth as butter" stream switching, play-list support, splitting the gst-plugins-bad module, Bluetooth plug-ins, and development tools/debugging improvements. For development tools/debugging there are hopes for better error reporting, tools collection, and ideas floating around about dot file dumps (dumping the GStreamer pipeline and also ideas for making it interactive).
Also be sure to read GStreamer 1.0 Is Looking To Finally Be Released Soon.
Add A Comment