----- Original Message ---- From: Vin Shelton <acs(a)xemacs.org&gt; To: Rick Rankin <rrankin1424-xemacs(a)yahoo.com&gt; Cc: xemacs-patches(a)xemacs.org; Dr. Volker Zell <dr.volker.zell(a)oracle.com&gt; Sent: Tuesday, January 24, 2006 2:24:01 PM Subject: Re: [PATCH 21.4] Xpm no-X builds fail on current cygwin > > Aha! It turns out it matters whether you run /bin/gcc or /usr/bin/gcc - > [ output elided ] > > I think my patch is good, because it supports both /bin/gcc and > /usr/bin/gcc :-) > > Rick and Volker - what do you think? > Well, it's nice to know what's going on. I verified on my system that I get the same output as you with /bin/gcc vs. /usr/bin/gcc. I went back and looked at the patch, but I think I need to look at the code in configure.ac. Basically, I'm not sure I understand why we need all this machinery to begin with. What all this evaluates to is '/usr/include', which is what it should evaluate to on every cygwin system, regardless of where cygwin is installed. The non-x version of XPM is in /usr/include/noX. It's not clear to me what all this extra machinery is buying us. If the header files aren't in those locations for some reason, it's not clear to me that any of this is going to work, anyway. In fact, the existing code didn't work when I had a home-grown version of gcc installed for a while. I had to force configure/make to use the version of gcc that is packaged with cygwin. Also, it appears from some of the surrounding patch text that the mingw headers are being included in a cygwin build, which would not be correct. Again, though, I need to look at this in more context which I won't be able to do until this evening at the earliest.