On 11/10/13 04:58, Landry Breuil wrote: > On Sun, Nov 10, 2013 at 04:55:29AM -0500, Stuart Cassoff wrote: >> On 11/10/13 04:39, Landry Breuil wrote: >>> On Sun, Nov 10, 2013 at 04:17:32AM -0500, Stuart Cassoff wrote: >>>> On 11/10/13 04:02, Landry Breuil wrote: >>>>> On Sun, Nov 10, 2013 at 02:42:22AM -0500, Stuart Cassoff wrote: >>>>>> Update to 1.15. >>>>>> Some stuff went upstream. >>>>>> Instead of doing all that ridiculous work in the Makefile, >>>>>> I've moved it all to Tcllib's installer itself. >>>>>> This should make future updates a lot less painful. >>>>>> I'll fill in MASTER_SITES before commiting. >>>>>> Distfile on cvs:~stu >>>>> >>>>> So upstream makes releases, but dont provide tarballs ? Or it's a >>>>> handrolled tarball from a git hash ? >>>>> >>>>> Landry >>>>> >>>> >>>> They make releases but there have been substantial changes since the last >>>> release. >>>> I keep pushing stuff upstream. There'd be many more patches if we used the >>>> release. >>>> For Tklib it'd be even worse, something like 40+ patches. >>>> >>>> Hand-rolled tarball from fossil. >>>> >>>> In the Makefile I should s/snapshot/checkin/. >>>> >>>> What about naming? Would tcllib-1.15.<checkinId>.tar.gz be too much? >>> >>> something like 1.15.2013xxxx to make it clear it's _post_ 1.15. >>> >>> Landry >>> >>> >>> >> >> Gr. Chekin id is wrong in the Makefile. >> >> Base version + shortened checkin id. Points to the exact spot that it came >> from. >> >> tcllib-1.15_fdbe5cb1bc.tar.gz > > Problem with that is that it's not a real version, so you cant use it as > a PKGNAME. Your call, use the checking hash for DISTNAME then handroll > PKGNAME with a .201311xx, or just use a date for both so you dont have > to futze PKGNAME. >
Ok so I'll go with tcllib-1.15.20131111.tar.gz (assuming tomorrow ;) or whatever date, and mention the checkin id in the Makefile. Stu