Re: Alternate automake output?

[William Robertson]

On 23 July 2002, Tom Tromey address@hidden wrote:
| >>>>> "William" == William Robertson <address@hidden> writes:
| 
| William> I've made the usual search through the docs and mailing list,
| William> but haven't been able to find an answer for this.  Is there
| William> any way to tell automake to generate an alternate output file
| William> instead of Makefile.in?
| 
| In theory this can be done.  In practice I don't know if I've ever
| even tried it.
| 
| But try writing something like:
| 
|     AC_OUTPUT(foo)
| 
| Then write `foo.am'.  This should cause automake to create `foo.in'.
| 
| I think that's all that is currently possible.  And it won't work
| properly in some situations, namely recursive make invocations.

Bummer.

| Why do you want to do this?

Ok, here's the situation.  I have a modules directory in which live a
bunch of packages which, by external mandate, must have the ability to
be compiled both separately and as a (set of) groups.  So, they each
have their own autoconf setup, the top-level configure.in uses
AC_CONFIG_SUBDIRS and some conditional logic, and everything is good.

Good, except that configuration runs are abysmally slow.  So, I'm
looking into ways of speeding things up.  One thought I had was to have
the top-level configure generate Makefiles for each package when
compiling modules as a group, using a separate Makefile.in like
AC_CONFIG_FILES([subdir/Makefile:subdir/Makefile.top.in]) or some such,
thus avoiding redundant configuration runs for each module while
retaining their build independence.

The sticking point is getting automake, when run from the top level, to
generate something other than a Makefile.in in each module directory, so
the module's local Makefile.in doesn't get clobbered.

I know this is a hack, but could automake play along with this, and
would this work?  Alternatively, is there a cleaner way to achieve this
goal?

-- 
| William Robertson | GPG:0x90A3BED4 | address@hidden |