On Thu, Dec 30, 2004 at 08:36:54AM -0600, Craig A. Berry wrote: > I just noticed a couple of other problems I haven't previously reported. > > 1.) The clean target renames the Makefile, so there is nothing left > with which to run the realclean target. This is contrary to the > comment inserted in the clean section and also contrary to my memory > of what we've always done. Unix has done that for as long as I can remember. VMS must have fell behind. Descrip.mms_old is left around if you wish to run the realclean target and realclean is smart enough to try for it in subdirectories in case they were also cleaned. Its possible it was just something to fix a GNU make issue as the comment in makefile() suggests but I don't see much harm in doing it everywhere and it removes a bunch of duplication. > 2.) MM_Unix::fixin doesn't seem to clean up properly on VMS. Here's > an example borrowed from the $(INST_SCRIPT)instmodsh target showing > that it successfully creates a new file with the .new extension but > does not rename it back to the original name: > > $ dir [.blib.script] > > Directory D0:[CRAIG.trunk.BLIB.SCRIPT] > > .exists;1 instmodsh.;1 > > Total of 2 files. > $ perl "-I[.lib]" "-MExtUtils::MY" -e "MY->fixin(shift)" [.blib.script]instmodsh > $ dir [.blib.script] > > Directory D0:[CRAIG.trunk.BLIB.SCRIPT] > > .exists;1 instmodsh.new;1 Does rename have the same issues on VMS as unlink? -- Michael G Schwern schwern@pobox.com http://www.pobox.com/~schwern/ Look! Defenseless babies!Thread Previous | Thread Next