Re: gcc_contrib bug

[Gerald Pfeifer <gerald at pfeifer dot com>]

Hi DJ,

I agree that an infinite loop isn't exactly helpful.  I believe the
original code for that aspect of contrib/gcc_update was contributed 
by Alexandre; perhaps he's got an idea?

Gerald

On Tue, 1 Apr 2008, DJ Delorie wrote:
> we use the following logic:
> 
> ... :
> 	@for f in $? $@; do test -f $$f || exit 0; done; \
> 	echo Touching $@...; \
> 	touch $@
> 
> We have the following chain of dependencies:
> 
> gcc/configure: gcc/configure.ac
> gcc/cstamp-h.in: gcc/configure.ac
> gcc/config.in: gcc/cstamp-h.in
> 
> In my particular build, for some reason, gcc/cstamp-h.in is missing
> but the other files are there.  I now have four gcc_update tasks
> running, trying desparately to figure out what to do.  It won't touch
> gcc/cstamp-h.in becuase of the @for line, but it always thinks it
> needs to update gcc/config.in because gcc/cstamp-h.in is missing.
> 
> Note that gcc/cstamp-in.h is NOT listed in $?, which is a list of
> newer prereqs - but not *missing* prereqs.  Perhaps we should use $^,
> which is a list of all prereqs?  That leaves us not touching
> gcc/config.in at all, but at least it doesn't hang.
> 
> At the very least, some sanity about how many times we permit
> ourselves to run that Makefile would be useful.  I mean, if it doesn't
> work in the first 345,600 iterations, it's not ever going to work.