On Wed, Apr 04, 2007 at 05:33:09PM -0400, Jason Harris wrote: > On Thu, Apr 05, 2007 at 03:20:11AM +0800, Rong-En Fan wrote: > > On Wed, Apr 04, 2007 at 03:10:38PM -0400, Jason Harris wrote: > > > > Rather than invoke dozens of needless processes, I simply pass > > > MD5_FILE=/dev/null to the "make fetch" in the makesum target. > > > (Note that this also removes the need to set NO_CHECKSUM=yes and > > > DISABLE_SIZE=yes in the same target, further simplfying things.) > > > > Note that it does not solve 'make fetch' when there is no MD5_FILE > > (this is another regression with latest bsd.port.mk). DISTINFO_DATA > > still needs to check existence of MD5_FILE. > > Remember, this is the "make makesum" step we're all (supposed to be) > talking about. That is, MD5_FILE is truncated to 0 bytes and is > added to in several steps. The regression, AFAICT, is that awk > is run when MD5_FILE doesn't exist (rather than empty/truncated), > returns non-zero, and kills the make.
I understand what you said in previous mail. I just want to point out the regression is not only restricted to 'make makesum' :-) Regards, Rong-En Fan
pgpH38DNp1Yuf.pgp
Description: PGP signature
