On Mon, Apr 8, 2013 at 6:45 AM, David Tardon <dtardon@xxxxxxxxxx> wrote:
On Sun, Apr 07, 2013 at 07:11:44AM -0400, Nico Kadel-Garcia wrote:You are totally missing the point. What I tried to convey is that
> > Why was Java 1.4 succeeded by Java 5? Why was ICU 4.8.1 succeeded by ICU
> 49.1? Why does systemd have version 197 instead of 1.9.7 or somesuch?
>
> If you look at the source code and the package names, Java wasn't really
> renumbered that badly, Java 1.4 was succeeded by Java 1.5, and Java 1.5 by
> Java 1.6. They're just marketed misleadingly. Sun's desicsion to screw up
> numbering this way is a reflection of when they tried to "SunOS 4" as
> "Solaris 2.5".
expectation that every upstream adheres to your favourite versioning
scheme is not based on reality. You cannot base an argument on it,
because the premise is already invalid. (Did I already say that in the
part that you removed?)
No, I'm trying to point out that if you actually look at the source code, the numbering is not that awful. It's when you try to number things based on marketing that confusion arises,and Sun was a major contributor to this problem.
And yes, there actually _are_ projects that use the usual X.Y.Z
versioning scheme, but _did not_ start with version 1.0.0. What comes
immediately to my mind is libreoffice (the first release was 3.3.0) or
libexttextcat (the first release was 3.2.0).
That's because it was a fork from another software package that was already at major version 3.x. And you don't have to backfill release numbers for completeness. Just stick with the GNU spec of x.y.z, x for major new releases, x.y for releases that add new features but will still run the older x release features, and x.y.z for minor releases that are compatible among all x.y numbers.
> And don't *get* me going on mod_perl numbering or CPAN versionAnother point for my case.
> numbering. (Whose bright idea was it to use floating point? Version 2.237
> is older than version 2.3 ?)
If your point were that some people deliberately ignored sensible and published standards and left the rest of the world to suck it up and deal, then yes. But it's the exception, rather than the rule, if you actually check the source code bundles.
-- devel mailing list devel@xxxxxxxxxxxxxxxxxxxxxxx https://admin.fedoraproject.org/mailman/listinfo/devel
