Re: Failure in test silent5.test with heirloom make

[Ralf Wildenhues]

* Stefano Lattarini wrote on Thu, Apr 22, 2010 at 02:25:08PM CEST:
> At Tuesday 20 April 2010, Ralf Wildenhues wrote:
> > Yeah, heirlooom make has a .l.o rule that triggers before our .l.c
> > and .c.o rule chains.
> > 
> > I guess this could be worked around by adding explicit rules (at
> > least that's what SUSv3 recommends),
> > maybe explicit dependencies without rules suffice.
> Yes, they does.  I hacked a bit the silent5.test to demonstrate this 
> (the diff is in the attached "patch").

That is indeed a good indication.

> However, I don't know enough 
> Automake internals to easily write a patch integrating this knowledge 
> into Automake core.

First step would be to write a new testsuite test to clearly expose this
issue; for example by adding a flag to yacc or lex invocation that won't
be added by the internal rules.  Silent mode is not relevant for this.

Then, look at automake.in:lang_yacc_target_hook.  It already computes
the name of the .c file.  The missing part is then to find out whether
the .$(OBJEXT) or .lo file is renamed (and thus will already be handled
by an explicit rule anyway), or whether it will be handled by one of the
suffix rules .c.$(OBJEXT) or .c.lo.  Only in the latter two cases we
should add an explicit dependency.

If you have trouble with the second part, the first would already help.

> >  I'm not sure we should spend time on this old make, though.
> If I venture to write a patch to solve this problem, will you
> consider it?

Yes, if it's not dangerous or regresses other situations.

Cheers,
Ralf