Conversation
| item = Item().findOne({'folderId': folder['_id'], 'meta.codec': 'h264',}) | ||
| if item: | ||
| video = Item().childFiles(item)[0] | ||
| if video: |
There was a problem hiding this comment.
Drop this condition. An item tagged with meta.codec should be an item with one file. If the child file is missing, that's effectively a database schema violation and a 500 error would be appropriate.
Alternatively, keep the if and add an else that throws a more verbose exception.
There was a problem hiding this comment.
I originally didn't have it in because I assumed that it should have a single file child, I reviewed and didn't know if forever (in terms of future updates) it would have a single file. I'll drop it.
There was a problem hiding this comment.
Yeah that's sort of the trouble with girder overall. When you base your entire app on freeform metadata inside folders and items, your schemas get expressed throughout the whole project as code.
By putting an if here you say "I have a union type" and then you have to hope that's consistent with all the other code that manipulates this data.
It scales pretty poorly. That was a large part of my motivation for writing the postprocessing endpoint: I wanted to get all that data-affirming code in one place and let that function be the source of truth for schema.
I guess I should add some logic in that function to verify the results of the processing pipelines. Like, after image or video conversion runs, postprocessing should go through and make sure all the metadata was set.
2 participants
Changes the video file endpoint to the
/fileendpoint. This endpoint properly supports streaming of files using the correct range headers. This should fix the issue where it takes an insanely long time to load and playback large videos. The previous version would try to download the entire file up to that one location in the space.Let me know if there is a better way or pattern to do this.
Fixes #273
Fixes the core of what we can reproduce for #171