From: Godmar Back (gback@cs.utah.edu)
Date: Mon Jan 18 1999 - 10:29:30 EST
>
> On Jan 17, 1999, Godmar Back <gback@cs.utah.edu> wrote:
>
> >> That's very odd... I can't reproduce this in my source tree. But I
> >> see your home dir is NFS-mounted, which may indicate clock drift.
>
> > Nope.
>
> > I'm building in /x/gback/transvirtual/obj and I have the source in
> > /x/gback/transvirtual/kaffe, both of which are local on marker.
> > My homedir simply has a symlink to /x/gback.
>
> Another possibility has just come to my mind: marker is *very* fast.
> Therefore, it may sometimes create files that depend on one another
> with identical timestamps. Next time `make' looks at them, it will
> rebuild the dependent file, just to be on the safe side.
>
Hmmm, but why hasn't this happened before?
When comparing timestamps, shouldn't you only compare the
time stamps of the .o files created by "make all" with the .c files
from which they were created? Those .c files do not change between the
time when "make all install" proceeds from the "all" to the
"install" target.
Of course, it could be that all creates some time stamp files to
indicate its success, and that install depends on this time stamp
files. If they are too young, what you describe could happen.
This would then be a bug, of course.
- Godmar
This archive was generated by hypermail 2b29 : Sat Sep 23 2000 - 19:57:44 EDT