> From: Junio C Hamano [mailto:gitster@xxxxxxxxx] > Sent: Friday, August 24, 2012 6:07 PM > To: Joachim Schmitz > Cc: git@xxxxxxxxxxxxxxx; Erik Faye-Lund > Subject: Re: [PATCH v2] Support non-WIN32 system lacking poll() while keeping the WIN32 part intact > > "Joachim Schmitz" <jojo@xxxxxxxxxxxxxxxxxx> writes: > > > There is a downside with this: In order to make use of it, in Makefile it > > adds "-Icompat/win32" to COMPAR_CFLAGS. This results in > > compat/win32/dirent.h to be found, rather than /usr/include/dirent.h. > > This should be fine for WIN32, but for everybody else may not. > > For HP NonStop in particular it results in a warning: > > > > }; > > ^ > > "... /compat/win32/dirent.h", line 17: warning(133): expected an identifier > > > > And this is because there it uses an unnamed union, which is a GCC extension > > (just like unnamed struct), but not part of C89/C99/POSIX. > > > > One possible solution might be to move compat/win32/poll.[ch] to compat/. > > I think that is the most sensible way to go, because poll.[ch] > > (1) has proven itself to be useful outside the context of win32, and > (2) the code is coming from gnulib anyway. > > I thought I already suggested going that route in my previous > review, but perhaps I forgot to do so. No, I believe you did, but I had forgotten about it. Could/should that be a 2nd patch? Or a v3 of this one? Different, but related question: would poll.[ch] be allowed to #include "git-compat-util.h"? Bye, Jojo -- To unsubscribe from this list: send the line "unsubscribe git" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html