* Stefano Lattarini wrote on Tue, Jun 22, 2010 at 01:15:04PM CEST: > At Monday 21 June 2010, Ralf Wildenhues wrote: > > I'm falling more and more behind on patches, it seems. > In this last period, it seems quite the opposite to me :-) > Thanks for your work!
You've done most of the actual work, giving me less reason to say no. ;-) > > Prefer `make -e' over `make VAR=VAL', to please > > maintainer-check. > ... are we sure we want to relax the tests just to please maintainer-check? > After all, those tests were working correctly with both GNU and BSD make, > and (mostly) with Solaris and Heirloom make That is only the case for those variable overrides for which there is no setting in the makefile. Solaris make doesn't get this right. What we could do instead is make use of AM_MAKEFLAGS to portably transport overrides to submakes. > (with failures due to those > make implementations bypassing the Automake-generated ".y => .c => .o" > rule chain in favor of their own ".y => .o" built-in rule, but that's > another story). That's another story altogether, let's discuss that in the thread that deals with it already. > I think it's time to make `maintainer-check' more configurable and > whitelist-friendly. No disagreement from me; but bear in mind that it should not get less powerful at detecting unportabilities. > And there's no need to remain anchored to grep and > sed: since we can assume that perl is available, we can as well use all > its power... I'm a big fan of grep, sed, and awk. I'll prefer the Posix tools whenever feasible. Cheers, Ralf
