https://gcc.gnu.org/bugzilla/show_bug.cgi?id=65142
--- Comment #10 from rguenther at suse dot de <rguenther at suse dot de> --- On Tue, 15 Sep 2015, redi at gcc dot gnu.org wrote: > https://gcc.gnu.org/bugzilla/show_bug.cgi?id=65142 > > --- Comment #8 from Jonathan Wakely <redi at gcc dot gnu.org> --- > (In reply to Richard Biener from comment #7) > > If the user controls how the random file is opened (non-blocking or > > blocking) > > They don't. > > > then the behavior (whether to re-try on EINTR or short reads) should be > > controlled by that choice. Starting to throw on users that don't expect > > that > > would be bad. > > The function is specified to throw on error by the standard. Well, the question is what is an "error" then. The need to wait (as you say we open blocking) isn't in my view. Getting EINTRed while waiting neither. Getting a fatal error from the read yes.
