mint release from a tag on github

[yarikoptic]

ATM it just lists them as tags, but not readily shows the most recent tag/release. So we should mint a proper release for tags we push.

[jwodder]

@yarikoptic What should be done about already-pushed tags? Should the script just create a release for every tag that lacks a release?

[yarikoptic]

ideally -- yes, create release for every tag matching the regex for our versions. Before that -- should be done in the order probably from older to newer. May be should not be done "all the time" but just as a helper command (release-already-existing-tags [dandisets]) which we could run once. Then we would just rely on new pushed tags to be accompanied with making a release for them.

[jwodder]

@yarikoptic

create release for every tag matching the regex for our versions.

Do you expect there to be tags that don't match the version regex?

---

I think always ensuring that all tags have releases would be easier than trying to figure out which tags were just pushed (unless you know an easy way to do that?).

[yarikoptic]

ATM I don't know of any tags which aren't release tags, but who knows -- we might add/push some later on. But ok, mint for every tag, although I would have limited to simple (not extended) semver ones (MAJOR.MINOR.PATCH) .

As for checking all tags upon every push - I think that would introduce unnecessary burden which would slow down otherwise possibly fast push which would not push any tag. But ok -- if you think that overhead is minimal, let's just always do that

[yarikoptic]

BTW, can't even quite get how/where do we push the tags since seems we use datalad's `Dataset.push` but that one seems to not bother pushing tags (unless git config sets what to push but we do not)

```shell ❯ python3 -c 'import datalad.api as dl, json; ds=dl.Dataset("."); res = ds.push(to="localhost", result_renderer=None); print(json.dumps(res, indent=2))' [ { "action": "publish", "refds": "/tmp/tttttt", "type": "dataset", "path": "/tmp/tttttt", "status": "notneeded", "target": "localhost", "refspec": "refs/heads/git-annex:refs/heads/git-annex", "operations": [ "uptodate" ], "hints": null, "message": "refs/heads/git-annex->localhost:refs/heads/git-annex [up to date]" }, { "action": "publish", "refds": "/tmp/tttttt", "type": "dataset", "path": "/tmp/tttttt", "status": "ok", "target": "localhost", "refspec": "refs/heads/master:refs/heads/master", "operations": [ "fast-forward" ], "hints": null, "message": "refs/heads/master->localhost:refs/heads/master c0fb8ec..06433db" } ] ❯ git push --tags localhost Enumerating objects: 1, done. Counting objects: 100% (1/1), done. Writing objects: 100% (1/1), 174 bytes | 174.00 KiB/s, done. Total 1 (delta 0), reused 0 (delta 0), pack-reused 0 To /tmp/tttttt-localhost * [new tag] 1234 -> 1234 * [new tag] tag2 -> tag2 ```

[jwodder]

@yarikoptic Tags are pushed here. Given this, creating releases will be easiest if a release is only created for a just-pushed tag. If you then want there to be a release-already-existing-tags command, should this be a subcommand of backups2datalad or a separate script?

[yarikoptic]

@yarikoptic Tags are pushed here. Given this, creating releases will be easiest if a release is only created for a just-pushed tag.

yeap -- great, let's add there.

If you then want there to be a release-already-existing-tags command, should this be a subcommand of backups2datalad or a separate script?

whatever is easiest/fastest -- I expect/hope to never use it again ;-)