Lukas Fleischer [off-list ref] writes:
Oh, and one more detail: I wonder why we still use fwrite(), now that we
know we can use xwrite() which guarantees atomicity. Is there a reason
for that?
The code (before squashing anyway) used to use fprintf() directly
bypasing the outbuf for unusual cases, and we didn't want to mix raw
file descriptor access with stdio access.
That no longer is the case, so I think we can switch to xwrite(),
and lose the last paragraph of the log message entirely.
Thanks.