On 2021-08-18 18:41, Bozhidar Batsov wrote:

> I'm a bit confused by the conversation so far. Can someone elaborate

> on "maintainers have explicitly pushed their tags to the ELPA repo"?

> I do tag all the releases of my packages, as that's a common (and

> good) practice, but I don't understand why would something like this

> be affecting ELPA negatively.

It won't.  Not unless you went out of your way to make it a problem.

Your "single project repository" is not the ELPA repo.  The ELPA repo

contains all of the packages in that archive.

This whole discussion only applies if you are manually pushing code

changes to the ELPA repo.  If your package is defined as an external

repo for ELPA's build processes to fetch automatically, then you aren't

pushing *anything* to the ELPA repo at all.

> Does it sync the tags from the remotes or what?

No, it doesn't, so tags can't be a problem if ELPA is fetching the

updates itself.

> In general I don't think that something like "stop tagging your

> releases upstream" is a good solution.

Keep tagging to your heart's content in your own repository.  Just

don't push those tags to the ELPA repo (which you would need to do

explicitly with options to the "git push" command).

> Adding a prefix to the tag name (e.g. the package name) also seems

> weird in the context of a single project repository.

Again, this was purely in the context of the multi-project ELPA repo.

If someone particularly wanted tags in the ELPA repo for a package,

then the tags would need to be namespaced with a prefix in order to

avoid potential clashes.

-Phil