Alexander Kolbasov wrote: >>>>>> "Darren" == Darren J Moffat <[EMAIL PROTECTED]> writes: > Darren J Moffat <[EMAIL PROTECTED]> wrote: > > Darren> DJM-1 Why has priv_const.c moved from generic makefiles to > $MACH ones? This is > Darren> just begging for it to get out of sync between intel and sparc. > Darren> I know it is a simple rule but can't the rule stay in some > Darren> common Makefile ? > > This common Makefile is included by everyone and, as a result, we > sometimes attempt to build the file when we should not. The priv_const.c > is the prerequisite that should be build up-front before we go and build > any parallel subdirs. With the new prerequisite targets we have common > prerequisites for x86 and sparc but no shared prerequisites. An > alternative would be to move this into uts/Makefile and just build it > before everything else. The difficulty is that we only want to do this > for certain targets which will produce ugly uts/Makefile rules. How > strong are your concerns here?
My concern is less with this particular case and more with the style precedent of duplicating targets it may appear to set. I'm okay with you integrating it as is if you add a comment that changes in the intel one should probably be duplicated in the sparc one and vice versa. > Darren> DJM-2 Why all the new @pwd added to rules ? What does this cost > ? and is it > Darren> worth what it gives us ? > > There are just a few pwd statements added here - all of them for the > lint when it generates global cross-check, so the cost is negligible and > it provides a bit better observability when we stare at the nightly > output and try to figure out what is is going on. What is your concern here? If the cost is negligible and it provides and observability benefit I have no concern. My concern was that it might not have been negligible. -- Darren J Moffat _______________________________________________ tools-discuss mailing list [email protected]
