Found a WEBM that has a different MD5 than gelbooru has.

gelbooru.com/index.php?page=post&s=view&id=4008117

ID: 4008117
MD5: 448769d3d0d4f00a3f1e24b602f2b797 (gelbooru one, taken from thumb URL, can be found with this)
MD5: 0af8b98d9699b45e2f003e2f80d9f52b (actual of the WEBM-file, 0 search results)

gelbooru.com/index.php?pa...ist&tags=id%3a4008117
gelbooru.com/index.php?pa...3d0d4f00a3f1e24b602f2b797
gelbooru.com/index.php?pa...d9699b45e2f003e2f80d9f52b

There might be more.

If I am not mistaken, that's the md5 of the original file that was uploaded, as it takes time for each video to convert.

Latest webms do not have md5 mismatch, so it seems that the original problem got fixed a long time ago, but this posts mismatch (and others before the fix happened) still exist. Shouldn't that be corrected? Because right now it's MD5 value is kinda useless.

When a user uploads a valid webm, it isn't converted, as it doesn't need to be. It's already the format we serve. If a user uploads a avi, mpeg, mkv, it'll be converted and the hash will be of the original file. We are preventing the same source video from being converted twice, as it's a waste of time to do that for a duplicate file, which would already exist on the site.

So I misunderstood there. It is indeed a good reason to store original md5, but why keep it as "file's md5"? Non-webm videos seem to be a special case, so why not make a special field or table for original values of converted videos? Because right now, if I have such video, I won't be able to find info about it on site by it's md5, so auto-tagging and sorting won't be applicable for them and that is my concern.

But I suppose keeping videos saved from gelbooru isn't a good idea to begin with. I'll have to rethink that.