On 02/03/15 07:20, Bernhard Voelker wrote: > On 02/26/2015 10:30 AM, Pádraig Brady wrote: >> Perhaps we should generalize this to --on-error to cater for all cases? > > maybe yes. We'll only have a problem if the OpenGroup would define > the behavior wrt. open() failures in a different direction than our > implementation ... which is then bound to the behavior upon write() > failures. I think it would be best to let POSIX clarify on this > first. WDYT?
It would be an extreme edge case that you'd want to ignore open() errors, while honoring write() errors. I'll send a patch to merge both under the --output-error option, rather than --write-error. thanks, Pádraig.
