Re: Automake violations of the gnu coding conventions

[Ralf Wildenhues]

* K. Richard Pixley wrote on Wed, Jun 20, 2007 at 08:21:55PM CEST:
> Ralf Wildenhues wrote:
> >Only if the source tree is read-only AND also the autotools ARE 
> >present, then I can see how you get a failure.
> That is my current situation, yes.  Automake is present and the source
> tree is read-only.

Richard, we seem to be miscommunicating.  You stated something different
a few messages up this thread (namely that automake is not present),
I've asked you for a copy and paste output of a make error, about three
times.  Now you claim that your current situation is a different one
than what you claimed three messages ago.  But we *still* do not get to
*one* exact reproducible scenario, and you *still* do not bring proof
that *that* is the actual failure that you are encountering.  There may
still be yet another bug that we are not aware of, and that I may be
able to see when seeing the error message, but now I am not; aside from
the fact that I would have caught the inconsistency in your description.

So really there is no way around you presenting me exact data.  If it
helps you, think of me as a stupid little kid that needs to be shown,
pointed to in the most obvious sense, but by actual code and by actual
error messages corresponding to exactly that code (just like the stuff I
showed to you a few messages earlier), not by talk about it.  Automake
is so full of subtleties that it is almost impossible to debug by talk.

If you have more than one scenario, feel free to do the same several
times, but you really have to do it at least once: if you ever want a
patch accepted in Automake (and we're several steps and some
controversial points-yet-to-be-discussed away from that) that changes
functionality then new testsuite tests that expose the issue(s) are a
fundamental requirement.

Regards,
Ralf