PushIOvHelper errors

[Russ Allbery]

Alex Kiernan <alexk at demon.net> writes:

> But when I was about to commit it, I set to thinking - if we've a write
> error and we don't cope with the retries, the client is going to be out
> of sync with us, so should we just call ExitWithStats()?

PushIOvHelper should be calling xwritev so that short writes are handled
internally.  Once it does that, the only way that it can fail is if we
really can't write to the remote host, at which point calling
ExitWithStats is correct, I think.

-- 
Russ Allbery (rra at stanford.edu)             <http://www.eyrie.org/~eagle/>

    Please send questions to the list rather than mailing me directly.
     <http://www.eyrie.org/~eagle/faqs/questions.html> explains why.