-
Notifications
You must be signed in to change notification settings - Fork 113
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Change how to format a data type with a record constructor #662
Conversation
)" This reverts commit a7f3b2a.
Unfortunately, squashing has two effects which make it undesirable:
|
I don't think this is a problem. We can still bisect to find a PR that introduced a bug and, if necessary, continue bisecting on the PR's branch to find the exact commit. It may be an extra, but tiny work. Rather, since all commits related to a PR are squashed, we can get a meaningful diff and easily access to the PR for the underlying discussion.
I'm surprised by this behavior; I expected all commits created on GitHub would automatically be GPG-signed. Actually I don't care even if a commit is not signed, though. Overall, I don't think squashing a PR has large problems. But if you are concerned about them, I don't have any objections to merge (not squash) PRs. |
Yeah, I think it is a bug in GitHub here. I am concerned about signed commits, is a step for supply-chain security. But we also have very little need to revert PRs, so let's go with squashing. |
After thinking a little, I've decided to specialize the case where a
data
or anewtype
declaration has only one record constructor. It breaks the consistency, but I don't think people will define multiple record constructors often.Fixes: #546
@mihaimaruseac By the way, could you merge PRs by squashing? It will make the commit log more readable. Of course, I can squash commits by myself, but it'll require force-pushing and I want to avoid it as I may make a mistake. Squashing-merging is easy and handy, I think.