Re: let's trace?

[Alexandre Duret-Lutz]

>>> "Tom" == Tom Tromey <address@hidden> writes:

[...]
 adl> I know of at least one diagnostic that the trace code
 adl> doesn't make: it doesn't warn that AC_CONFIG_HEADER must
 adl> be rewriten as AM_CONFIG_HEADER.
[...]
 Tom> We need AM_CONFIG_HEADER to let us make the stamp files.
 Tom> I think we can do this using m4 tricks, just like we do
 Tom> with other things (dependencies) from init.m4.  That seems
 Tom> most robust to me, it means the user has one less AM_/AC_
 Tom> split to remember, and it means one less error mode in
 Tom> automake.

This sounds better to me too.

 adl> * lib/am/header-vars.am

 Tom> This .am file is special.  See define_standard_variables.
 Tom> Maybe we need more tests to test for this stuff.  As I
 Tom> recall this special code lets us use `+=' on built-in
 Tom> variables.  Anyway, removing definitions from this file
 Tom> might not be fully correct.

Holy crap!  I was completly ignorant of this.  However after
reading this function I think it's ok to remove all AC_SUBSTed
variables from header-vars.am (what I did) since
&define_standard_variables also defines all the AC_SUBSTed
variables.

 Tom> I'm all for moving to traces as soon as possible.  It is
 Tom> clearly the way forward and we want the maximum amount of
 Tom> time to find the inevitable bugs.

I'll commit this momently.
-- 
Alexandre Duret-Lutz