swissspidy
commented
3 weeks ago From the MDN documentation:
Each time you call createObjectURL(), a new object URL is created, even if you've already created one for the same object. Each of these must be released by calling URL.revokeObjectURL() when you no longer need them.
Browsers will release object URLs automatically when the document is unloaded; however, for optimal performance and memory usage, if there are safe times when you can explicitly unload them, you should do so.
There are two more places outside the upload queue where object URLs are created:
- Media recording, after recording is finished and before the file is inserted/uploaded
- Block transforms, when dropping multiple files, causing the individual blocks to upload them separately
- Aside: they won't have the same
batchIdthis way.
- Aside: they won't have the same
Gutenberg handles the revoking in the uploadMedia callback:
Blob URLs are currently created in lots of places, but not explicitly recoked again with
revokeObjectURL. Ideally that should happen as soon as possible, but only when s blob URL isn‘t used anywhere anymore. So the safest time to do it would be after removing an item from the queue.Need to keep track of all blob URLs per item so they can be revoked upon removal.