Bug #119123 for version: alpha version ordering is broken

Sun Dec 04 17:20:59 2016 VDB [...] cpan.org - Ticket created

Subject:

alpha version ordering is broken

Consider a simple test: use version; my $v = version->parse( "v0.11.2_4" ); print $v->normal, "\n"; If version < 0.9913 the test prints v0.11.2_4 Internally it is represented as FOUR-component array [ 0, 11, 2, 4 ]. Semantics of Alpha version is not well-defined in the documentation, but I assume this is proper result, because it gives order: v0.11.2 < v0.11.2_4 < v.0.11.3. If version >= 0.0013, the test prints v0.11.24 Internally it is represented as THREE-component array [ 0, 11, 24 ]. It breaks version ordering. Current version order is quite useless: v0.11.2 < v0.11.3 < ... < v0.11.23 < v.0.11.2_4.

Mon Dec 05 14:14:54 2016 DBOOK [...] cpan.org - Correspondence added

On Sun Dec 04 17:20:59 2016, VDB wrote: Show quoted text

> Consider a simple test: > > use version; > my $v = version->parse( "v0.11.2_4" ); > print $v->normal, "\n"; > > If version < 0.9913 the test prints > > v0.11.2_4 > > Internally it is represented as FOUR-component array [ 0, 11, 2, 4 ]. > Semantics of Alpha version is not well-defined in the documentation, > but I assume this is proper result, because it gives order: v0.11.2 < > v0.11.2_4 < v.0.11.3. > > If version >= 0.0013, the test prints > > v0.11.24 > > Internally it is represented as THREE-component array [ 0, 11, 24 ]. > > It breaks version ordering. Current version order is quite useless: > v0.11.2 < v0.11.3 < ... < v0.11.23 < v.0.11.2_4.

This was an intentional change to make it more consistent. See https://github.com/Leont/version-bad/pull/1 . In version tuples the underscore is no longer considered a separator but merely an indication of alpha version. It is recommended not to use underscores with version tuples, but instead to upload your distribution with -TRIAL in the name to prevent indexing (which is done automatically by dzil release --trial).

Mon Dec 05 14:14:55 2016 The RT System itself - Status changed from 'new' to 'open'

Mon Dec 05 14:17:30 2016 DBOOK [...] cpan.org - Cc DBOOK added

Tue Dec 06 15:38:51 2016 ether [...] cpan.org - Cc ETHER added

Wed Dec 07 16:31:42 2016 VDB [...] cpan.org - Correspondence added

I would expect such a breaking change is documented in Changes. However, version's Changes is automatically generated and useless stuff: it is not recognized by metacpan (which usually shows changes in the current release at the top of distribution page, but not for version), for human reader changes provides lots of noise (branch created/merged/tag added) with no value.

Wed Dec 07 17:15:20 2016 DBOOK [...] cpan.org - Correspondence added

On Wed Dec 07 16:31:42 2016, VDB wrote: Show quoted text

> I would expect such a breaking change is documented in Changes. > > However, version's Changes is automatically generated and useless > stuff: it is not recognized by metacpan (which usually shows changes > in the current release at the top of distribution page, but not for > version), for human reader changes provides lots of noise (branch > created/merged/tag added) with no value.

Agreed. Unfortunately this distribution has a nonstandard layout where the useful changes are documented in README, which should probably be in the module documentation somewhere if not fixed. This change can be found here: https://metacpan.org/source/JPEACOCK/version-0.9917/README#L38

Sat Dec 10 17:47:58 2016 VDB [...] cpan.org - Correspondence added

On Wed Dec 07 17:15:20 2016, DBOOK wrote: Show quoted text

> Agreed. Unfortunately this distribution has a nonstandard layout where > the useful changes are documented in README, which should probably be > in the module documentation somewhere if not fixed. This change can be > found here: https://metacpan.org/source/JPEACOCK/version- > 0.9917/README#L38

I see. Why don't you rename README to Changes? Reformatting it to meet Changes requirements does not look too difficult.

Sat Apr 15 06:58:50 2017 jpeacock [...] cpan.org - Correspondence added

Fix will be included in 0.9918 (soonish)

Sat Apr 15 06:58:51 2017 jpeacock [...] cpan.org - Status changed from 'open' to 'resolved'