Hi Michael,
On Tue, 8 Dec 2020, Michael W Powell wrote:
> On Wednesday, December 9, 2020 at 12:06:33 AM UTC-5 Michael W Powell wrote:
>
> > What is this? And how is it not a version bump? For the third time,
> > actually... i.e. (1), (2), (3)... Thank you.
> >
>
> The naming convention is consistent at least, but it is a bit odd to me.
> What is the .3 number in the convention, please? What do you call that?
> Besides major (2.), minor (.29), patch (.2)...
Git for Windows is a fork of Git, and therefore follows Git's versions. So
whenever a new Git version comes out (the next is most likely v2.30.0, for
example), we release a new Git for Windows version whose version number is
derived from Git's (in the case of v2.30.0, the tag will be called
v2.30.0.windows.1).
Sometimes, Git for Windows needs to release intermediate versions that do
not follow a release by the Git project. Most recently, those were
critical security bug fixes in Git LFS and in Git Credential Manager Core.
Those did not require any updates in the Git project, but in Git for
Windows, because Git for Windows actually bundles binaries and ships
related software (Git LFS and Git Credential Manager Core, for example).
The tags corresponding to these fixes are called v2.29.0.windows.2 and
v2.29.0.windows.3, respectively.
Technically, the most important changes were not even in the
git-for-windows/git repository, so the tagged revisions do not
_themselves_ have the fixes, but we require those tags so that we can have
proper releases up at
https://github.com/git-for-windows/git/releases.
Since we're still based on Git v2.29.0 in both of those cases, the version
number therefore still starts with that string, but then adds the number
after the `.windows.` part of the tag in parentheses.
By that naming convention, Git for Windows v2.29.0 should actually have
been called v2.29.0(1), but for convenience, I simply drop the `(1)`.
There will never be a `(0)` because we will always have patches on top of
upstream Git.
Does that clarify things?
Ciao,
Johannes