Amitai Schleier

Summarizing and reminding ourselves: the path forward here is emphatically _not_ syncdir. That would be a relatively complicated fix. Since we're the actively maintained upstream, we instead will directly add...

Before syncdir became Unlicensed, I worked up a proof-of-concept branch integrating its logic directly. Now that [syncdir is Unlicensed](https://untroubled.org/syncdir/), I'll rework that branch and perhaps propose a PR.

Related, perhaps to be solved separately: > As it turns out, macOS cheats. On Linux, fsync() will both flush writes to the drive, and ask it to flush its write...

Yes, I'd feel much better if we could directly test the real `qmail-inject` we're about to install. Trouble is, before we've installed, `qmail-inject` hits `die_chdir()` -- and if we _have_...

I've made the failing test fail the build, so the checks should look uniformly bad now :-) In an attempt to avoid - hand-duplicating an entire rule just so I...

Also, the filter no longer drops a sentinel-file turd to indicate the bug. Instead it uses the customerror API from #195. Perfect fit :-)

Wonderful. I’ve linked to this from

Is this ready for re-review?

This should be on our roadmap, no question. I'm worried that if we make a change like this now, before we've implemented most of the features people are currently applying...

Added the regression-test idea as #17.