2026/03/22

Sunday, March 22, 2026 (#81)

Saturday Sunday Monday posts: prev < this > next

References 2026/03/21 ‎ (← links) 2026/03/23 ‎ (← links)

• 09:53 The solution I've come up with at this point is to store the Runner-object-pointer alongside the OClient() and OServer() object-pointers, in QORunner() ("QO" instead of just "O" because it won't necessarily be set), in the Engine object.
• 10:17 ...and now we're back to the point of "execution gets stuck somewhere" (probably waiting for input when it shouldn't be, or something).
• 11:02 For one thing, NewSchemaList() was opening its own Runner connection, which it shouldn't do, but I'm not sure this actually made any difference. Fixed that, but still getting hung. Onward...
• 11:06 The sticking-point is somewhere in DoInOutLoop() (who could have foreseen)
• 11:15 Yep, it's stuck waiting for an input Stream which supposedly has not reached feof()...
  • ...so presumably we can't trust feof() here, but we also have to trust it elsewhere...
  • ...so presumably we need two different classes, one for each of those two different contexts...
  • ...but trying to figure out for sure what any given Stream is, is nontrivial (could probably do it from existing info if I had more braincells, BUT)
  • ...so I think I have to make each Stream-object require a name-string, even if it's only ever used for debugging. Meh.
• 21:48 Is it possible that I'm opening the process and then calling DoInOutLoop() before sending any SQL for it to execute?
  • This does not appear to be the case.