On Oct 8 10:03, Eric Blake wrote: > On 10/8/18 3:24 AM, Houder wrote: > > grep fails on Cygwin because lseek() on Cygwin fails to recognize > > that it is applied to a FIFO. As result of that, it returns EINVAL > > in errno, where it should return ESPIPE. > > > > Receiving the wrong value in errno forces reset() (in src/grep.c) > > to return false, upon which grep fails. > > > > Before v2.27 of grep, a call to S_ISREG(st->st_mode) in reset() > > prevented the call of lseek() and made reset() return true. > > > > The call to S_ISREG() has been removed in v2.27 of grep. > > > > The solution would be either to correct Cygwin's executive or to > > s/executive/dll/ > > > insert a Cygwin-specific kludge in grep (in reset() ). > > I see no reason to do a one-off kludge to my build of grep (since that does > not scale - every other app that also makes decisions based on errno values > would have to make the same kludge), compared to just fixing cygwin1.dll for > everyone. > > But thanks for isolating the problem!
I pushed a patch. Thanks for tracking this down, Henri! Corinna -- Corinna Vinschen Please, send mails regarding Cygwin to Cygwin Maintainer cygwin AT cygwin DOT com Red Hat
signature.asc
Description: PGP signature
