On Thursday 25 April 2002 12:40 pm, Jean-Marc Lasgouttes wrote: > >>>>> "Angus" == Angus Leeming <[EMAIL PROTECTED]> writes: > > Angus> Ususally, this results in config.h being unchanged, but not > Angus> this time. It is totally altered, mostly by a re-ordering of > Angus> the variables which makes it hard to see if anything has really > Angus> changed. > > I changed an AC_CHECK_FUNCS call (for mkstemp in particular) so that > it is inside AC_LANG_C (exactly the thing I told you not to do for > your own use :). I do not see how it could have changed config.h, > though. > > JMarc > > > Angus> Clearly, the ones of interest are those that are no longer > Angus> defined. Can anybody put my mind at rest, or should I > Angus> investigate further. (What changed in the configure stuff > Angus> anyway...) > > It may be interesting to investigate. Are yu sure you did not change > your autoconf version in the meantime?
That probably explains why the "safe" and new versions have different orders. I note that autoconf 2.5 results in the XPM_H_LOCATION and FORMS_H_LOCATION being at the top of config.h for some reason... I downgraded to autoconf 2.13 a couple of weeks ago since you told me that that was a good idea. It desn't explain why all those HAVE_STRUCT_STAT_XXX etc defines have gone, but I ran "cvs diff configure.in -r XXX" all the way back to XXX=1.104 and could see nothing obvious. As they aren't actually used, it's not a problem; I just got worried. > JMarc Angus
