Add close-issue-reason option #764
Conversation
README.md
Outdated
| @@ -80,6 +80,7 @@ Every argument is optional. | |||
| | [ignore-updates](#ignore-updates) | Any update (update/comment) can reset the stale idle time on the issues/PRs | `false` | | |||
| | [ignore-issue-updates](#ignore-issue-updates) | Override [ignore-updates](#ignore-updates) for issues only | | | |||
| | [ignore-pr-updates](#ignore-pr-updates) | Override [ignore-updates](#ignore-updates) for PRs only | | | |||
| | [close-as-not-planned](#close-as-not-planned) | Use the "not planned" close reason for issues | | | |||
There was a problem hiding this comment.
IMO because this only affects issues (PR closed state is either unmerged or merged), and because of the existence of options like close-issue-message, close-issue-label, etc. This option should be close-issue-<something>.
Also, right now it's a boolean (either not planned or completed), but that can change if GitHub ever introduces a 3rd closed state. That's why I suggested in #744 (comment) to have a close-issue-reason option that takes a string.
But don't make any changes just for me 🙂
Let's wait and see what the maintainers think.
There was a problem hiding this comment.
yeah I think I like the close-issue-reason improvement, which makes us a little more resilient/future proof to any new states
There was a problem hiding this comment.
Would you want to validate against a known set of values, or just pass it through to the API verbatim?
There was a problem hiding this comment.
I think validating against known values seems like a good idea. I haven't explored what happens if you were to pass state_reason: foo, so it's possible the API throws out bad values, but it's very possible it might also throw a 400.
So I think throwing out any values that aren't in the list ["closed", "not_planned"] makes sense.
There was a problem hiding this comment.
If I pass an invalid reason, including an empty string, I get 422 Unprocessable Entity:
{
"message": "Please specify a valid state reason.",
"documentation_url": "https://docs.github.com/rest/reference/issues#update-an-issue"
}There was a problem hiding this comment.
ah ok, I think in that case, it makes sense to check if it's one of the 2 valid values. If not, just fallback to the default. What do you think?
There was a problem hiding this comment.
Sounds good, it just means that if GitHub introduces a new value in the future, we'll have to push another update to this action.
There was a problem hiding this comment.
yep, understood. IMHO that's preferable to potentially breaking the action in a potentially confusing way
There was a problem hiding this comment.
@luketomlinson changes made, please take another look when you get a chance. Thanks!
jtbandes
changed the title
|
Great, thanks for reviewing/merging! Are you planning to make a new release soon as well? |
**User-Facing Changes** None **Description** Uses actions/stale#764 to close issues as not planned instead of completed
**User-Facing Changes** None **Description** Uses actions/stale#764 to close issues as not planned instead of completed
**User-Facing Changes** None **Description** Uses actions/stale#764 to close issues as not planned instead of completed
Changes
Adds an option
close-issue-reason, which can be used to close stale issues as "not planned" (a relatively new GitHub feature).I tested the
state_reason: "not_planned"option by making manual calls to the REST API (unfortunately it is not currently listed in the REST API documentation).Context
Resolves #744