magit-tag-release: also read message for --sign/--local-user

[corecode]

git tag requires a message for all "heavy" tags that have an associated tag object. --annotate, --sign, and --local-user= all create tag objects. Make magit-tag-release read a message in all of these cases.

[tarsius]

git tag requires a message for all "heavy" tags that have an associated tag object. --annotate, --sign, and --local-user= all create tag objects.

If git tag is called with one of these arguments but without -m, then it invokes $EDITOR for the user to write the tag message. As for commit messages, this is convenient when we want to provide more text than fits on a single line.

• If you invoke magit-tag-release with one of the sign arguments (but without --annotate), then a buffer pops up.

• If you invoke magit-tag-release with --annotate (regardless of whether a signing argument is also used), then the user is instead required to provide the message in the minibuffer.

This difference in behavior is intentional; the user can decide how they want to provide the message. However, I also believe I only considered the case where the user signs tags ("because tags should always be signed"). I.e., it is currently not possible to not sign the tag and also use the minibuffer to write the message.

One solution that comes to mind, is to change magit-tag-release so that it does not use the minibuffer if --annotate is enabled. (Instead of, as you propose, to extend that special handling of that argument to the signing arguments as you propose.)

The special handling would not be removed, instead the -m argument should be added to the menu, and if that is enabled, then magit-tag-release would read a message in the minibuffer (instead of using a proper buffer, or not adding an annotation at all, depending on the other arguments).

• Users could then enable -m, if they want to use the minibuffer.
• If they want to use a proper buffer, they could instead enable --annotate (or a signing argument).
• If they don't want an annotation, then they should keep all arguments disabled.

We would of course also have to consider how that affects magit-tag-create.

[corecode]

magit-tag-release currently uses magit-run-git-async and doesn't set EDITOR, that's why I am submitting this PR.

I figured I would retain the spirit of magit-tag-release and read from the minibuf. It felt like magit-tag-create is the function that will do the whole process with editor popup and no prepared tag/message, while magit-tag-release optionally prompts for a message. Both --sign and --local-user imply --annotate, so it feels like magit-tag-release should act accordingly.

[tarsius]

magit-tag-release currently uses magit-run-git-async and doesn't set EDITOR,

I've fixed that.

[corecode]

Do you intend magit-tag-release to act like magit-tag-create if --sign is used (but not --annotate)? I did expect it to act as if --annotate was selected (because --sign does imply it).

[tarsius]

I intend to make these commands consistent with each other, and with git tag (if appropriate). The topic isn't in my working memory right now, so I don't know what that entails. I'll work on it once I don't have much more important matters to attend to.