* Michael G Schwern <schw...@pobox.com> [2008-01-13 17:50]: > Peter da Silva wrote: > > I tried that for a while, but even *I* couldn't keep track of > > what version numbers meant, and it was my own project. > > > > If the current version is 1.5.4 and the guy's running 1.5.2 > > that tells me more than if the current version's 20070620 and > > the guy's running 19990114. > > It tells you that it's eight years old, and that's concrete > information. You know that eight years is a damn long time in > software years and it's a flag that you should probably look > into it. It's not much, but it's something. > > What does 1.5.4 vs 1.5.2 really tell you?
Everything, if *I* am the one who *minted* those version numbers, as per the case that Peter mentioned. And it tells me just as much if I am one of many others who have some knowledge about what happened in a particular version. It’s much easier for the maintainer to remember 2.4.10 as unusual than 20010923. > Oooh, and then there's the odd/even alpha/release fun. Now is > 1.4.6 newer than 1.5.1? Could be! Can you safely use 1.5.1? Who > knows? Haha, have fun figuring it out, sucker! A user sends me a bug report for 20010923. Hmm, was that a release in the stable branch or in the devel branch? Who remembers? Crap, let me dig through the metadata records to find out. > The reality is in the mind of the person who did the release. Exactly. If I am the person who is doing the release, version numbers are much more memorable than dates. > You can't resolve that difference of visibility and > expectations between author and user, so I'm leaving the game. Pick your poison. All approaches suck. Regards, -- Aristotle Pagaltzis // <http://plasmasturm.org/>
