egcs-19981012: libf2c installation problem

[Billinghurst, David (RTD)]

Yes, by rebuild I mean recompiles everything.

Start from scratch and do full configure and make in $builddir
cd $builddir/mips-sgi-irix6.2/libf2c
make -n 		# shows nothing to be done
make			# everything is recompiled
make -n  		# shows nothing to be done
make			# everything is recompiled again

I am pretty sure the problem is the interaction between
g2c.h and Makefile in $builddir/mips-sgi-irix6.2/libf2c

g2c.h depends on Makefile, among other things.
g2c is (re)generated by ./config.status, which then (seems to) regenerate
Makefile too ;-(
Certainly the timestamp on Makefile is consistently 1 second after g2c.h

There are a number of comments in the Makefile about g2c.h. so it obviously
isn't simple. 
Perhaps we could 
	get ./config.status generate Makefile first, then g2c.h, or
	touch g2c.h after generating Makefile

> -----Original Message-----
> From:	Dave Love [SMTP:d.love@dl.ac.uk]
> Sent:	Tuesday, October 20, 1998 2:07 AM
> To:	Billinghurst, David (RTD)
> Cc:	'egcs@cygnus.com'
> Subject:	Re: egcs-19981012: libf2c installation problem
> 
> >>>>> "DB" == Billinghurst, David (RTD)
> <David.Billinghurst@riotinto.com.au> writes:
> 
>  DB> Compiled and checked as usual "make; make -k check"
>  DB> Then "make install" rebuilds libf2c
> 
> Assuming `rebuilds' means actually recompiles things rather than just
> recursing and doing nothing: that can happen if you don't start with a
> clean build directory (for instance), since re-running configure
> doesn't update timestamps of generated files if the files don't
> change.  It's not clear to me what's the right thing to do; zapping
> all the generated files initially in the make rules is easy enough,
> but not by the book.
> 
>  DB> Cannot add further details without rebuilding whole distribution.
> 
> If it still wants to rebuild You should be able to figure out the
> reason from `make -n -d' in $builddir/mips-sgi-irix6.2/libf2c.  If, as
> implied above, the second build brought everything up-to-date I don't
> know why and would need more debugging info.