2026/04/07

Tuesday, April 7, 2026 (#97)

Monday Tuesday Wednesday posts: prev < this > next

References 2026/04/06 (← links) 2026/04/08 (← links)

• 08:27 {Trying to figure out why performance_schema gets stuck in a loop} is running into the problem of {getting the diagnostic information from {where it is available} to {where it can be included in the readout-panel}}.
• 08:46 ...and {trying to figure out {how to improve the situation}} is running into the fact that I use two different Updater objects in distinctly different ways: one checks the output, the other updates the screen.
  • Maybe if I look at what's happening in the output-checker, I'll get lucky and it will turn out that's where the problem is.
• 09:31 The export process doesn't use an output-checker, so nope.
  • That kinda leaves IsRunning() as the likely culprit, I think?
• 09:43 Thinking through the whole "why doesn't this work now even though it used to" thing:
  • Was I sending the command differently?
    • I don't think so; the dump command is a one-shot thing, which means it opens the session, which is equivalent to how it was being invoked before.
  • I was getting error messages back then, and I think they were in a separate channel. How did that work?
    • I guess that's a thing to check into. Maybe the messages are holding things up now.
• 10:01 Apparently, the error-messages were, indeed, keeping the process open even after the main output was done -- exports seem to be working again 🎉
  • Now I just need to figure out why the messages aren't being indicated on-screen (which presumably means they aren't being processed by the Messages handler)...
• 12:30
  • ✅ Messages now being correctly indicated on-screen.
  • ✅ All configs updated to latest format.
  • 🔎 Schema-listing broken for remote MySQL (though still working for remote MariaDB).
    • This is unsurprising, given the quirky I/O differences between the two.
• 13:42 Added the --unbuffered option to the mysql command. After some bug-fixing in the Messages clades, it now retrieves a schema-list from cloud5 (and nothing else seems rebroken).
• 21:25 Oops -- I did in fact re-break local-proc, but it was an easy fix. (Current task is trying, once again, to implement schema imports (in case that wasn't obvious).)
• 22:03 ...and we've gotten to the point where there are no errors, but nothing seems to be happening.
  • Correction: it does create the schema if not found. That's a start.
  • ...but then it really does get stuck, and a 10-second time-out trips in the Conveyer.