F77 makefile again.

craig@jcb-sc.com craig@jcb-sc.com
Wed Jun 30 15:43:00 GMT 1999


>> and we basically never get any proper bug reports showing failures
>> during parallel make.
>
>Gee, I see them every now and then.  I fix most of them I see.

You're fixing bugs in g77 makefiles?  Great, thanks!  If you know of
other complete bug reports against g77 makefiles, please point me
to them.

>> (The occasional bug report we do get just doesn't tell us enough to
>> help us understand the problem.)
>
>Gee, I haven't seen many that are incomplete.

Hmm, I have had several on file, e.g. just partial patches saying
things like "this is needed to make XYZ work" with no explanation,
no log file showing XYZ *not* working, etc.  I've gotten to where
I just delete these, especially after repeated unheeded requests to
re-submit as proper bug reports.

I've occasionally tried applying the recommended patch, only to
discover *it* was broken, and in fact there are often comments
in the area clearly explaining that such a change would break
the g77 makefiles.  But these went unread, or at least un-commented-upon,
by the person claiming his patch somehow would fixed things.  (Sometimes
they might have fixed one thing, but they broke others.)

The end result of this sad state of affairs is that the very person,
or persons, who refuses to read and follow the incredibly clear
documentation on how to read a gcc/g77 bug report, also goes around
telling everyone the g77 makefiles are "broken" because we g77 people
supposedly won't take the trouble to read the docs on `make' and
simply "get it right" in the first place.

(I've read the docs on `make', yet it is still the case that I don't
write perfect makefiles.  Similarly, I've read the docs on C, and
on Fortran, yet, amazingly enough, despite being considered by some
to be even an *expert* on those topics, I've never managed to write
a perfect Fortran compiler in C.  In both cases, I rely nearly exclusively
on decent bug reports.  Fortunately, my g77 users are quite understanding
of this -- if they can't formulate a decent bug report, due to a lack
of time/resources, they don't blame *me* for the bug not getting fixed.)

So I've decided to stop wasting time with g77 bug reports from the people
who persistently refuse to write proper bug reports but continue to blame
me (and my g77 co-workers) for not fixing the bugs, except in the rare
cases where I can clearly see my mistake and fix it without worrying
about breaking something else.

In the meantime, I might try `make -j3' as you suggested, and perhaps
also use a `make' option to have the dependencies resolve in reverse order,
if one exists, or create one myself (if that's easy enough) for my
own use, to help find bugs in g77's (and perhaps gcc's as well) makefiles
by doing things "backwards".

        tq vm, (burley)



More information about the Gcc mailing list