[erlang-questions] Version numbering scheme change and the implication / Re: [ANN] Erlang/OTP 17.0-rc1 has been released.

[Vlad Dumitrescu]

Hi!

I assume that the philosophy behind the version numbering of OTP hasn't
changed, only the format of it. However, a specification of how to
interpret version changes would be very helpful. As Anthony stated, it is
important how similar it is to semantic versioning.

* Will version changes in applications propagate to the OTP level, like for
semantic versioning?

* If yes above: OTP contains besides the core applications even tools and
libs that are loosely related. Will all applications be treated the same,
for example will a new major version of, say, eunit require updating the
major version of OTP?

* Will parts beyond the normal x.y.z be treated like in SemVer? I.e.
version X-rc1 sorts before X. Will there be "+something" parts too?

best regards,
Vlad



On Sat, Feb 1, 2014 at 7:09 PM, Ignas Vyšniauskas <baliulia@REDACTED>wrote:

> First of all, excited to hear about the RC!
>
> Anyway, I agree with everyone that the version numbering change is a bit
> fishy.
>
> As already mentioned, this is a major backwards incompatible change for
> everything that tries to parse the release version.
> Secondly, as Anthony pointed out, it is indeed not Semantic Versioning
> at all if it's anything like "17.0-rc1".
>
> Perhaps a middle ground solution would be to e.g. stick with something
> like.
>
>     R<MAJOR>[A|B]<MINOR>-<PATCH>-<OTHER>
>
> with, say the <PATCH> and <OTHER> parts being optional.
>
> In that way you can increment the versions using semantic versioning,
> but still stay as backwards compatible as possible (e.g. simple parsers
> that parse based on lexicographic order or only look at the major
> version will stay happy).
>
> --
> Ignas
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://erlang.org/pipermail/erlang-questions/attachments/20140203/f7164168/attachment.htm>