I know this is the proper way of doing it according to semantic versioning.
I was afraid of populating the Metacello config with lots of #version21:
#version211: #version212, etc.
But I also should be able to load a version with the bug, so I guess this
is way.

But... how do I load the latest "released" version of 2.1 ?

Ej:
I have:
- 2.1.3 (released)
- 2.2.4 (released)
- 2.3... (development)

If I want to load the latest released version of 2.1, how do I do it
without knowing if the latest is 2.1.3 or 2.1.15?

Regards!


Esteban A. Maringolo


2014-08-08 10:09 GMT-03:00 Esteban Lorenzano <esteba...@gmail.com>:

> well… my advice is to use semantic versioning (http://semver.org).
>
> in essence:
> - a package commit is just a commit. The number is a commit number (bah..
> is a little more complicated because the commit info is author-number, but
> you got the idea).
> - metacello configurations work as distribution artifacts, so they should
> have an unique and unrepeatable number (unless you are using git and
> baselines, in which case it changes). So you should not change the old
> version. Instead, you should add a patch number.
> In your case:
>
> 2.1
> will become:
> 2.1.1
>
> and then you can move the #stable symbolic version to 2.1.1
>
> Esteban
>
> On 08 Aug 2014, at 14:53, Esteban A. Maringolo <emaring...@gmail.com>
> wrote:
>
> > Hi,
> >
> > How do you manage released versions updates in Metacello?
> >
> > E.g.
> >
> > I have a version 2.1, with its spec with a dependency to
> 'PackageA.11.mcz'.
> > Later I find a bug which requires a new version of PackageA.11.mcz,
> everything goes ok as in the development/bleedingEdge version, which
> doesn't specify the file name, but how should I "backport" the change?
> >
> > What I am doing now is changing the version spec for '2.1' and updating
> the reference to 'PackageA.11.mcz' to 'PackageA.12.mcz', but I'm afraid
> this isn't the proper way of doing this.
> >
> > How do you manage this?
> >
> > In git I'd merge the commit in the 2.1 version branch, and then the CI
> that depends on such branch would incorporate the change.
> >
> > Regards!
> >
> > Esteban A. Maringolo
>
>
>

Reply via email to