Michael Jennings wrote:
I agree with the other comment to your post, obviously, but apart from that: It's not necessarily about version numbers "like that", either. I invented a rather far-fetched versioning scheme that obviously ought to be replaced by something else just for the sake of the argument, but there could equally well be a question of having to changing from one versioning scheme that on its own may be seen as perfectly sane to a different, equally sane one. And there is more than one; using "<major>.<minor>.<patchlevel>" is not the only "right" way, I think. It has some advantages, including the fact that everyone is familiar with it, but I can think of some arguments against it, too, notably that what exactly constitutes a major change and what is a minor one, is anybodys guess, or rather, probably just a political decision. Personally I used a date in the format "YYYYMMDD" for quite some time, but then I had to change to that other convention after all...On Monday, 28 February 2005, at 13:09:11 (+0100), Toralf Lund wrote:
Then how on Earth do you put everything right, if for some reason
someone has chosen the wrong versioning scheme for earlier releases?
By doing it right, forcing people to manually upgrade whilst directing [ ... ]
If they released a package with a version number like that, yes. It's
called "Natural Selection." Perhaps you're familiar with it.... :-)
What I'm trying to say is just that Maximum RPM misses the point somewhat when it talks about versioning schemes that require epoch updates on every release, and concludes based on this that epoch is bad. Yes, such versioning is bad, but the way I've always seen it, the scenario in question was not the one the epoch mechanism was designed for. I've assumed EPOCH is just intended as a "more major" version than VERSION, i.e. a value you increment if, and only if, you change the versioning scheme.
- Toralf
_______________________________________________ Rpm-list mailing list Rpm-list@xxxxxxxxxx https://www.redhat.com/mailman/listinfo/rpm-list
