> > some systems define EILSEQ in wchar.h rather than errno.h > > (I am not sure whether I had a list of some such systems at some > > point, or not). > > Thanks, could you please dig that up, or let us know from memory > as best you can? I'd like to add this info to the documentation.
I searched the internet, and unfortunately, the only kind of clue I found was this comment from libiconv's eilseq.m4 which, I think, was really the only clue that I had seen: # The EILSEQ errno value ought to be defined in <errno.h>, according to # ISO C 99 and POSIX. But some systems (like SunOS 4) don't define it, # and some systems (like BSD/OS) define it in <wchar.h> not <errno.h>. I checked the following systems: ia64-hpux 11.31, x86-windows (cygwin and MinGW), ppc-aix 5.3, sparc-solaris 2.8, x86-solaris 2.10, x86_64-darwin 10.8, FreeBSD 9.0 and GNU/Linux in general (x86, x86_64, and ppc). All of them defined EILSEQ in errno.h (or in a file eventually included by errno.h). The reference to BSD/OS seems a little obsolete at this point, as wikipedia reports that this proprietary system was discontinued in 2004. I regret now expressing my concerns about EILSEQ being defined in wchar.h, rather than errno.h, as I feel like I may have made you do extra work for no real reason. Perhaps you might prefer using your first patch instead of the second one. It potentially leaves a problem unfixed; but on the other hand, very few systems would be impacted, and it leaving this open like this would allow us to know for sure if the problem ever came up. I have a feeling this would only be relevant to older systems anyway. > +2012-11-05 Paul Eggert <egg...@cs.ucla.edu> > + > + errno: port to LynxOS 178 2.2.2 > + Problem reported by Joel Brobecker in > + <http://lists.gnu.org/archive/html/bug-gnulib/2012-10/msg00088.html>. > + * doc/posix-headers/errno.texi (errno.h): Document this. > + * lib/errno.in.h (EILSEQ, GNULIB_defined_EILSEQ): New macros, > + defined if @EILSEQ_HIDDEN@ or if EILSEQ is not defined. > + * lib/strerror-override.c, lib/strerror-override.h (strerror_override): > + Supply a string for EILSEQ. > + * m4/errno_h.m4 (gl_HEADER_ERRNO_H): Check for EILSEQ. > + (gl_REPLACE_ERRNO_VALUE): New parameter $2, to specify what > + to do when the errno value is hidden. All uses changed. > + * modules/errno (errno.h): Substitute EILSEQ_HIDDEN and EILSEQ_VALUE. I've tested both of your patches on native x86_64-linux (system errno.h used, no errno.h generated), as well as when cross-building GDBserver for ppc-lynx178 (errno.h generated, with EILSEQ defined to our arbitrary value). Thanks a lot for all of you help, Paul - I really appreciate it. -- Joel
