If we choose to do it over existing HTTP APIs, we should probably wait until NGP is released as it will likely have a more comprehensive API than scratch.mit.edu and apiclone currently have.
This reminded me of the registration checks, so I made a HAR with those. It's worth noting that, at least I think, all of the domains it allowed through the email check have valid MX records.
Sorry I've been lazy regarding this, but I came across a rather interesting error message (also 200 OK for an error is crazy)
I suppose they aren't as useful as I thought then.
(sorry if this is getting off-topic but) it seems as if RSS feeds for topics are not paginated at all, are meant to be machine-readable, and contain the HTML of all posts (stored separately from...
I wrote some code to calculate the post count, [...document.querySelectorAll('td.tc3')].map(node => BigInt(node.innerText)).reduce((a, c) => a + c, 0n). As of writing this it returned 7786017n.
I tried to clean the comment hars mentioned earlier, hopefully I won't get h4x0r'd from uploading these.
Opened some (perhaps somewhat relevant) tracking issues for some features, perhaps I'll work on them when less lazy.
Judging from an archive.org snapshot I found and the live site, it seems as if scratchr2 and scratch-api respectively handle empty search queries very differently at the very least.
Made some comment hars (I kept getting ratelimited after making a reply to a reply, that's kept too), I'll clean them later.
I heard they're replacing the forums; do they intend to archive the old ones? If not we should probably do something about that (perhaps we could get Archive Team involved?)
Any luck with running the migrations (python3 manage.py migrate I believe)?