From: John Cowan <xxxxxx@ccil.org>
Date: Wednesday, July 10, 2019 2:56 PM

On Wed, Jul 10, 2019 at 2:51 PM Lassi Kortela <xxxxxx@lassi.io> wrote:

Does anyone lurking here know what's the story with sync() now that SSDs
are gradually taking over?

Given that disks, disk controllers, and disk array controllers all have big RAM buffers and report writes as done long before they actually are complete, I'd say: let's flush both sync-file and sync-file-system from SRFI 170.  That disposes of the question of which is the most useful (or least broken) scsh implementation.

This form of cutting the Gordian Knot works for me (insert even more Official caveats about Linux's fsync).  It could someday be the remit of a SRFI or library.

I'll remove them in my work-in-progress pull request.

- Harold