On Tue, 7 Jul 2026 at 16:52, Heikki Linnakangas <[email protected]> wrote: > Yeah, I guess that works.. But It's a little unsatisfactory to have to > poll, where we didn't poll before. I started to go down the rabbit hole, > here are the options I can think of:
Honestly, I like the fallback to 1-second polling quite a bit better than any of the options you proposed. The benefit of your options (afaict) is making sure a very rare 1 second delay in a cancel for pg_dump/pg_restore cannot happen on most platforms. That seems like a rather insignificant benefit. But to achieve that the code on Windows (and depending on the option other platforms) is now different, which makes testing any future changes to the code much harder for anyone who doesn't have Windows. I don't think the benefit is worth that downside in this instance. I guess my general thoughts on this is that I would like the behaviour of Windows and Unix to become closer, not further apart (unless there are significant benefits to doing so).
