On Wed, 2015-10-07 at 16:57 +0200, Mark Wielaard wrote: > So I changed the code again, now we just ignore all errors, except for > ENOSPC since that is the only real thing we are interested in. I pushed this variant to master now.