On 26 Jun 2008, at 21:52, Danek Duvall wrote: > On Thu, Jun 26, 2008 at 09:44:46PM +0100, Chris Ridd wrote: > >>> Each number between the dots (new fortune cookie reading?) is >>> treated >>> as an integer, so leading zeros are dropped. >> >> That seems like it might be a bit too simplistic, but maybe having a >> leading zero inside a version string is asking for trouble anyway. > > I'd agree with the second half of that. >>
:-) >>> It is also quite important to be able to compare the versions of two >> packages for ordering purposes. > > So if you have a version that can't be represented by a sequence of > integers, you'll need to encode it somehow into such a sequence, and > provide the unencoded version in a package attribute. Trying to > come up > with a generic scheme by which we can order xx.0c, -pre1, -rc-3, > -wild-weasel, etc, is something that we've decided to punt on, and > we'll be > leaving the encoding up to individual package maintainers. SUNWopenssl has that sort of problem - it isn't obvious from the packaging whether it is 0.9.8f, or 0.9.8g, or whatever. So your idea would work as long as the version in the FMRI were sufficiently well documented as only being there for the purposes of ordering and comparisons. Does a real_version package attribute exist yet? Cheers, Chris _______________________________________________ pkg-discuss mailing list [email protected] http://mail.opensolaris.org/mailman/listinfo/pkg-discuss
