| >>>>> "Harlan" == Harlan Stenn <[EMAIL PROTECTED]> writes:
| Harlan> However, if I move the lines:
| Harlan> ANSI2KNR = ../util/ansi2knr
| Harlan> ../util/ansi2knr:
| Harlan> cd ../util && $(MAKE) $(AM_MAKEFLAGS) ansi2knr
| Harlan> from their current position in the Makefile (late, after the
| Harlan> targets that refer to them) to a position "earlier" in the
| Harlan> Makefile, it works.
|
| Yeah, that's evil. I can't claim to understand it.
| Could you try this patch?
| (I just wrote it; I haven't tested it yet myself.)
|
| 2001-05-19 Tom Tromey <[EMAIL PROTECTED]>
|
| * automake.in (generate_makefile): Run handle_compile before
| handle_languages.
|
| Tom
|
| Index: automake.in
| ===================================================================
| RCS file: /cvs/automake/automake/automake.in,v
| retrieving revision 1.1126
| diff -u -r1.1126 automake.in
| --- automake.in 2001/05/18 17:43:46 1.1126
| +++ automake.in 2001/05/19 21:25:38
| @@ -1238,9 +1238,12 @@
| &handle_programs;
| &handle_scripts;
|
| + # This must run first so that the ANSI2KNR definition is generated
| + # before it is used by the _.c rules. This works around an
| + # obscure bug.
| + &handle_compile;
| # This must be run after all the sources are scanned.
| &handle_languages;
| - &handle_compile;
|
| # Re-init SOURCES. FIXME: other code shouldn't depend on this
| # (but currently does).
Hm, are you sure there are no dependencies between those two guys? I
don't feel good here. (Checking...) Hm, seems to be OK.
I really don't like having to do that. Could the fact that we are
using a macro for the dependency be responsible? Harlan, what happens
if you replace the macro invocation with its value?
