Filter results for “gnu”:
Languages
Toughness
Just bugs labeled...
Expand all
Collapse all
-
I want to use box.net or my ownbox instance as storage backend.
-
· link
I ran EAGER_CELERY and killed the server while transcoding media. The MediaEntries remain in "processing" state but the jobs will not restart/continue. Would it be possible to scan the database on GMG startup looking for entries in "processing" state and pass them to celery for processing? I noticed this now because I have a never-clearing queue of "processing" videos in my processing panel.
-
At least for image-based media it would be nice to present slide show views for tag/user/collection-based series of media. We could make use of http://www.justinbarkhuff.com/lab/lightbox_slideshow/ or something similar and. I guess this is mostly a template job as the backend should kind of exist. Use case: I create a collection of my last holiday pictures or family event and want to share those with my friends and family for convenient viewing.
-
· link
There is a new table for data used while (or at the end of) processing. It's currently only used for the callback url. We should use it for more things. So I propose to move various columns from `MediaEntry` into `ProcessingMetaData`. Possible columns: * transcoding_progress * queued_task_id * queued_media_file (*) * fail_error (*) * fail_metadata (*) After a successful processing, none of these are needed. After a failed processing, those with (*) might be needed for various reasons. We should rethink that part maybe anyway.
-
· link
Hi, I found the closure-compiler jar in extlib/leaflet/lib/closure-compiler/compiler.jar. Is closure a hard dependency or could mediagoblin be built without it? closure is not packaged for Debian so it would make it even harder to get mediagoblin in Debian. How do you plan to create a source only tarball if you commit .jar files to git? I'd propose to rely on closure-compiler being available and probably provide a two liner shell script to download and sha1sum the .jar file. Regards, Thomas Koch
-
It appears that recordings added are converted to WebM (Vorbis + VP8 inside the modified Matroska container.) Can you please add support to instead encode into Ogg Vorbis + Theora? Every browser that has built-in support for WebM also has support for Ogg Vorbis + Theora but the reverse is not true, giving Ogg Vorbis + Theora greater browser compatibility. Add to this that the Java Cortado applet [0] can be used as a fallback method means that browsers as far back as Netscape 4 can play these files. [0] http://www.theora.org/cortado/
-
· link
Sometimes a video or audio file is uploaded in a format used by GMG for publishing files, but GMG transcodes the file anyway. That makes sense for security and possibly privacy, but sometimes it’s more of a waste of resources. Proposal: an option to let files be used without transcoding. * Use that way only video files up to certain dimensions (compared to the target dimensions), maybe also considering the file size (files with big video dimensions may still be OK if the file size is small). * Transcode only the data that is not in the target format. Thus an Ogg/Vorbis file becomes WebM/Vorbis without transcoding. * For security reasons, the option should probably be disabled server-side by default. Maybe (in the future) enable it for particular users or user groups.
-
Currently, we store a media's license through it's UNICODE URL, this means lots of duplication. A patch to set a user's preferred default license also stores the preference as Unicode. Would it make sense to have a table containing the available licenses (this would let us include more things like "Human readable help text" etc per license easily and a media's license would refer to it via a ForeignKey (and so would a user's default license choice). Less duplication and license customization would happen by fudging with the table (so could be done by web admin interface or plugin or whatever).
-
{{{#!rstPlease add a way to change UI language, overriding Accept-Language.E.g. a drop-down and/or an URL piece.}}}
-
· link
We should have a unit test that instantiates the tables at their first, pre-migration state (basically right after the move to SQL, commit ce29c140ed90fe0e19ed1b68ca8f7cf27aa03e8e) and runs all migrations, then creates a new database and instantiates the models at their current state, and sees if they match. If they don't, the test should fail. Note: we can't test for constraints in sqlalchemy, but we should otherwise be able to test to moderately general migration sanity. (Bonus points if it adds a user and a couple of media entries with appropriate media files, etc, but not required to close this bug.)
