It's seems to be better to sort the tags that way, no need to worry about the format of the tag name.

If we decide to go this way, I still need to fix the tests

I'm a little reluctant to rely on pygit2's C variable there, I'd advice to use isinstance() instead

I have updated the code to use isinstance. I have also updated the tests, it takes around 20s (time to create different commit)

I am not quite sure we can test the case when we get a commit object and not a tag with the pygit2 API.

Why the sleep here?

ah yes, so that there is a time difference b/w the tags

Was it not working here?

I've just removed test case 3 which was adding version with alpha and beta to the tags created in case 2.
Now alpha and beta are in case 2 by default.
I have also shorten the tag lists so that the test doesn't take too long to run

Case 1 is still here

But after the repo contains something, not on an empty repo

True, not sure if it changes something regarding the tags, but I guess I can a another case Case 1 - Empty repo no tags
Case 2 - repo with commit and no tags .

:thumbsup: for me, makes sure we don't break on projects w/ empty repos :)

Changes made to test cases

Looks good to me, thanks!