Niels Thykier <[email protected]> writes: > I disagree that we should blindly assume that a "variable"-target as a > wildcard. Consider Lintian's d/rules that has:
> $(profiles): $(allchecks) private/generate-profiles.pl > LINTIAN_ROOT='.' private/generate-profiles.pl > This is completely unrelated to build-arch and build-indep, but its > presence would (as I understand you) make Lintian assume that > build-(arch|indep) is now present. Yup. Mostly I would have made that choice because it's a lot simpler and I don't think there are very many packages that use variables for rule targets, and most of the ones that do have a fairly sophisticated build system. But... > The alternative is to track variables and understand make. In theory > this very quickly becomes hard (to do right). However, if we assume > that people, who would use variables this way, are a bit sane, we can > get away with a far simpler approach. > Attached is a patch that does very simple (and in some cases even > incorrect[1]) tracking of variables and their values (ignoring tons of > rules). I still suspect it will not cause a not of false-negatives and > should reduce the number of false-positives. ...if you're willing to do the work, I'm certainly okay with this. :) We haven't gotten a lot of complaints about this yet, so if handling the simple cases along these lines makes people happy, that's an even better solution. -- Russ Allbery ([email protected]) <http://www.eyrie.org/~eagle/> -- To UNSUBSCRIBE, email to [email protected] with a subject of "unsubscribe". Trouble? Contact [email protected] Archive: http://lists.debian.org/[email protected]
