Skia minor version tracks milestone release branch

Skia minor version tracks milestone release branch

Also: version embedded into generated PDF document.

GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1781673002

[hal.canary, 2016-03-09 16:23:16]

Description was changed from

==========
Skia minor version tracks milestone release branch

Also: version embedded into generated PDF document.
==========

to

==========
Skia minor version tracks milestone release branch

Also: version embedded into generated PDF document.
GOLD_TRYBOT_URL=
https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&is...
==========

[hal.canary, 2016-03-09 16:24:06]

Description was changed from

==========
Skia minor version tracks milestone release branch

Also: version embedded into generated PDF document.
GOLD_TRYBOT_URL=
https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&is...
==========

to

==========
Skia minor version tracks milestone release branch

Also: version embedded into generated PDF document.

GOLD_TRYBOT_URL=
https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&is...
==========

[hal.canary, 2016-03-09 16:24:06]

halcanary@google.com changed reviewers:
+ reed@google.com

[hal.canary, 2016-03-09 16:25:03]

ptal

[reed1, 2016-03-09 16:37:37]

reed@google.com changed reviewers:
+ bungeman@google.com, jcgregorio@google.com, mtklein@google.com

[reed1, 2016-03-09 16:37:38]

+ben/mike/joe

[mtklein, 2016-03-09 16:48:25]

I don't think major.minor.patch fits what we do.

I would prefer to delete SKIA_VERSION_MAJOR, SKIA_VERSION_MINOR, and
SKIA_VERSION_PATCH, and if necessary add SKIA_MILESTONE.

[jcgregorio, 2016-03-09 16:54:46]

On 2016/03/09 at 16:37:38, reed wrote:
> +ben/mike/joe

This seems better than leaving it at 0, but:

  1. The usual standard is that the minor version number bumps only when
backward compatible changes are made, and the major changes when non-backward
compatible changes are made. We don't really stick to that for the C++ code as I
understand it.
  2. Would this version number be accessible to the C API?
  3. If we decide to go with this, we should also update the branch process to
include updating the minor version number.

My personal preference would be that we only support the C API as our "public"
way to interface with Skia, i.e. only our close customers use the C++ API, and
only the C API should be versioned, but only in the way described above.

[hal.canary, 2016-03-09 16:55:42]

On 2016/03/09 16:48:25, mtklein wrote:
> I don't think major.minor.patch fits what we do.
> 
> I would prefer to delete SKIA_VERSION_MAJOR, SKIA_VERSION_MINOR, and
> SKIA_VERSION_PATCH, and if necessary add SKIA_MILESTONE.

done (minus the deleting anything).

[mtklein, 2016-03-09 17:06:52]

> My personal preference would be that we only support the C API as our "public"
> way to interface with Skia, i.e. only our close customers use the C++ API, and
> only the C API should be versioned, but only in the way described above.

This sgtm.