From: John C. <jo...@st...> - 2002-09-10 13:54:27
|
I have just this minute got the brand new, shiny MinGW-2.0.0-3, and I will give it a go with that, after I manage to get the CVS sources. After that, we'll see... John. -----Original Message----- From: Earnie Boyd [mailto:ear...@ya...] Sent: 10 September 2002 12:38 To: John Cronin Cc: min...@li...; mak...@gn... Subject: Re: [Mingw-users] Make configuration for Win32. John Cronin wrote: > > To make things clear about bcopy: > > make.h checks for whether HAVE_MEMMOVE is defined and whether the > symbol bcopy is NOT defined (ie a simple symbol that the cpp would pick > up, not an autoconf generated one). If both of these are true, it > defines bcopy to memmove. The problem with this was, as mentioned in > a response to my query by pds (solely on the make-w32 list for those > not subscribed) that memmove had slipped out of the AC_CHECK_FUNCS > line in configure.in for 3.80rc1. I am assured it has been put back. > So the CVS version should handle this successfully. I'm glad to hear that. > About WINDOWS32: > I'm dropping the __WINDOWS32__ idea. It was just a thought about what MinGW team could do to help with the RMS desire. BTW, thank you Paul, for the explanation of the win terminology. > In my attempts to make make compile OOB for all current platforms and > Mingw32/MSYS, I rely on > > if test "$build_os" == "mingw32"; then > > (please point out if I should use target_os instead - but this isn't > checked for) > > to AC_DEFINE(WINDOWS32, 1, [Define if ...]) > and AM_CONDITIONAL(MINGW, true) > > else > > AM_CONDITIONAL(MINGW, false) > > fi > > which then has the makefile add in the w32 subdirectory to the build. > > This seems a simple fix, and I see no need to complicate matters with > another macro instead of WINDOWS32. > It seems that John has "stepped up to the plate" to get the w32 portions autoconfisticated properly. Can you carry forward with this John? Earnie. |