Hey folks,
You may have noticed we took about 2.5 hours of downtime this morning for...
Hey folks,
You may have noticed we took about 2.5 hours of downtime this morning for a
performance improvement.
Previously we memoized the number of upvotes and downvotes on both the stories and comments tables.
Many queries sorted by score, which they had to calculate per-row with upvotes - downvotes.
No queries directly used upvotes, so I replaced it by memoizing score instead.
The migration for this locked the tables, so we took downtime - that’s now completed. @355E3B and I are squishing have squished some small followup bugs; please file github issues or let us know in chat if you see anything odd.
And then as long as finishing that migration meant running it five times on my machine to iron out bugs, I picked up some other features we’ve been figuring out for a long while.
I’ve redesigned the tags page with categories to help make sense of it.
This was prompted by @355E3B’s analysis of our tags and the page’s
longstanding design issues.
I’m leaving the issue open because there’s still room for improvement.
Perhaps use our newly added svg graphing library to add sparklines showing usage?
I know categories are a really tempting bikeshed, so let me emphasize that categories and tags are not intended to be an epistemologically complete ontology of the entire field of computing.
They make the page look nice.
Maybe, like domains, they’ll slowly grow functionality, but let’s not overthink them too quickly.
I’ve revamped the comment flagging UI to further emphasize that they’re for alerting mods when a user doesn’t want to engage and curb their abuse for punishing disagreement.
Visually, flagging comments now looks like flagging stories with a text link rather than an down-pointing arrow, and flagging a comment collapses the tree.
Nothing on the site refers to “downvoting” any more.
I’ve also expanded the about page with material on flags and topicality compiled from
many meta stories and subthreads
about
topicality and the
purpose
of
flags.
These changes also address a recent thread on whether the warning about being heavily flagged
by trying to
improve
the
signal of flags so that granular warnings aren’t necessary.
(These links are not exhaustive.)
I’ve also deactivated the ‘Incorrect’ flag reason that was mostly used to disagree with people rather than indicate serious problems.
I hope this is a roughly accurate collection of community norms and expectations.
As usual for changes, these are not the final version we’ll ever have.
If you see ways to improve these things, I appreciate help identifying what’s wrong, proposing solutions, and implementing them.
Finally, last month I asked kyle if he was still interested in being a moderator, as it had been a long time since he was active.
He replied that he’s not not active online much anymore and not interested in continuing as a mod.
A few days later he deactivated a bunch of online accounts, including this one.
He leaves with our thanks for volunteering and best wishes.
Peter
