At 6:39 PM +0200 1/3/02, Jarkko Hietaniemi wrote: >On Wed, Jan 02, 2002 at 05:24:24PM -0600, Craig A. Berry wrote: >> At 03:43 AM 1/2/2002 +0200, Jarkko Hietaniemi wrote: >> > I went for even more extensive >> > rewrite of the logic, we'll see how it works. >> >> Not at all for me. There were no ppport.h or ppport_h files supplied with >> the distribution and no actions specified to build them. The following even >> more extensive rewrite works fine for me but has only been tested on VMS. > >grumble, grumble. I can't apply the below any more since it conflicts >with Nick's and Andreas' patches... Understood. I'll reformulate it when I can get to it (later today or tomorrow). >Surely we should be able to craft make rules that work across platforms >and satisfy the following: I believe mine does these things but will check the resulting makefile before I resubmit. In general I tried to increase use of standard arguments to WriteMakeFile and decrease weird things in the postamble. For example, MakeMaker already generates the line: $(O_FILES): $(H_FILES) and O_FILES is constructed from the list of C files, so 1b below is already taken care of if the "H" and "C" arguments are supplied to WriteMakeFile. The only real portability concern I had was whether double quotes around the -I options to perl would work everywhere. Perhaps I should use $(PERLRUNINST) instead of $(PERL). >(1) (1a) PPPort.o depends on PPPort.c and ppport.h > (1b) module[23].o depend on ppport.h > (1c) ppport.h depends on ppport_h.PL > (1d) PPPort.c depends on PPPort.xs >(2) no unnecessary recompiles (which means no unnecessary touching of files) >(3) the ppport.h is removed on clean -- ____________________________________________ Craig A. Berry mailto:craigberry@mac.com "Literary critics usually know what they're talking about. Even if they're wrong." -- Perl creator Larry WallThread Previous