Re: Serious (semi-)infinite loop

[Tom Tromey]

>>>>> "Tim" == Tim Van Holder <address@hidden> writes:

Tim> Running automake on libffi as part of the gcc tree, yields about
Tim> a zillion occurrences of

Thanks.
I agree this is a problem.

Tim> automake: Makefile.am: warning: automake does not support libffi_la_SOURCES
Tim> being defined conditionally

Interesting.

Tim>   libffi_la_SOURCES (User, where = 111) =
Tim>   {
Tim>     ALPHA_TRUE => $(libffi_la_common_SOURCES) $(TARGET_SRC_ALPHA)
Tim>     ARM_TRUE => $(libffi_la_common_SOURCES) $(TARGET_SRC_ARM)
Tim>     IA64_TRUE => $(libffi_la_common_SOURCES) $(TARGET_SRC_IA64)
Tim>     M68K_TRUE => $(libffi_la_common_SOURCES) $(TARGET_SRC_M68K)
Tim>     MIPS_GCC_TRUE => $(libffi_la_common_SOURCES) $(TARGET_SRC_MIPS_GCC)
Tim>     MIPS_SGI_TRUE => $(libffi_la_common_SOURCES) $(TARGET_SRC_MIPS_SGI)
Tim>     POWERPC_TRUE => $(libffi_la_common_SOURCES) $(TARGET_SRC_POWERPC)
Tim>     SPARC_TRUE => $(libffi_la_common_SOURCES) $(TARGET_SRC_SPARC)
Tim>     X86_TRUE => $(libffi_la_common_SOURCES) $(TARGET_SRC_X86)
Tim>     X86_DJGPP_TRUE => $(libffi_la_common_SOURCES) $(TARGET_SRC_X86_DJGPP)
Tim>     X86_WIN32_TRUE => $(libffi_la_common_SOURCES) $(TARGET_SRC_X86_WIN32)
Tim>   }

This is debug output which I've gotten rid of.

Tim> interspersed with messages like
Tim> automake: Makefile.am: object `sysv.lo' created by `src/m68k/sysv.S' and
Tim> `src/x86/sysv.S'

Also interesting.

The libffi Makefile is, and always has been, weird.  I've struggled
with it many times.  However, it does seem like a reasonable approach
to the problem.  It probably ought to work.

It might not be possible to make it work as-is with automake 1.5 :-(.
That's because of the combinatorial explosion problem discussed on the
main automake list.


Oh, yeah -- I just remembered that this Makefile.am relies on a hacked
automake 1.4 and does some other bad things.  For instance it has both
`src/mips/n32.S' and `src/mips/n32.s'.  These aren't even
distinguishable on Windows.

So making it work with 1.5 isn't a priority.
Adding `subdir-objects' helps, but doesn't really solve the problem.

Anyway, I'll definitely look into this and see what, if anything, can
be done for 1.5.  At the very least I'll document the supported way to
conditionally compile sources.

Tom