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

[Tuncer Ayaz]

On Fri, Feb 14, 2014 at 1:43 AM, Rickard Green wrote:
>>> When branching out, we add ".1" at the end of <X>.<Y>.<Z>, unless
>>> this version number has already been used. If it has already been
>>> used, we search for an unused version number by adding more and more
>>> ".0" between the version we are branching from, and the ".1" that we
>>> add at the end. For example, 17.0.1.1, 17.0.1.0.1, 17.0.1.0.0.1, and
>>> 17.0.1.0.0.0.1 are all versions of modifications based on version
>>> 17.0.1.
>>
>> I can follow the rest of your email, but can you provide some real
>> life examples for the case of inserting 0 upon branching?
>
> The version numbers above would be used if we need to publish four
> separate fixes and:
> - each one of them needs to be based on 17.0.1
> - 17.0.2 has already been published
> - each fix is only allowed to include specific changes

So, 'separate' in this case means 'independent' patches, and
therefore:

- 17.0.1.0.1 is 17.0.1 + 1st separate patch after 17.0.2 release

- 17.0.1.0.0.1 is 17.0.1 + 2nd separate patch released next

- 17.0.1.1 is 17.0.1 + 1st 'normal' patch

- 17.0.1.2 is 17.0.1 + 2nd 'normal' patch

Is that correct?

> Another scenario could be that we need to publish a sequence of four
> fixes and:
> - the first fix needs to be based on 17.0.1
> - 17.0.2 has already been published
> In this case we would use versions 17.0.1.1, 17.0.1.2, 17.0.1.3, and
> 17.0.1.4.

That makes sense.

> From time to time we have requirements like this, however, not to
> nearly as extreme as having to branch four times from the same
> version.

Sure, as long as as you don't priorize backport-ability over
refactoring :).