Refusing to use PRs

Posted Jan 13, 2026 9:53 UTC (Tue) by donald.buczek (subscriber, #112892)
In reply to: Refusing to use PRs by alx.manpages
Parent article: Evans: A data model for Git (and other docs updates)

> I use it even to document my github PRs, in projects I'm forced to use github. See the 'Revisions' in this comment:
<https://github.com/shadow-maint/shadow/pull/1471#issue-37...> ('git rd' is an alias I use: 'rd = range-diff master..@{u} master@{u}..HEAD').

Hmm, I don't know. You've come up with a personal standard that is more of a hindrance than a help to a potential reviewer. How is anyone supposed to know what it means? Where is revision “v1”? Why are there suddenly “b” and “c” variants in ‘v5’ but no “a”? Which version was or is now in the PR branch? In terms of content, there is free text under the versions, which is good, but then there is a privately defined git command (“git rd”) that has no clear meaning for anyone else and a list of commits where it is also unclear what they should be used for.

You might be better off sticking to the standard of the respective project instead of your personal one.

to post comments

Refusing to use PRs

Posted Jan 13, 2026 12:04 UTC (Tue) by alx.manpages (subscriber, #145117) [Link]

> > I use it even to document my github PRs, in projects I'm forced to use github. See the 'Revisions' in this comment: <https://github.com/shadow-maint/shadow/pull/1471#issue-37...> ('git rd' is an alias I use: 'rd = range-diff master..@{u} master@{u}..HEAD').
>
> Hmm, I don't know. You've come up with a personal standard that is more of a hindrance than a help to a potential reviewer. How is anyone supposed to know what it means?

That is a project where I contribute often and I help co-maintain. Maintainers know what to expect.

> Where is revision “v1”?

What I'm showing are range-diffs, so that if you've reviewed a previous revision, you know what changed for the next one.

v1 would compare to /dev/null, but that wouldn't be useful, since you're new to the patch set, so you need to review the entire PR at whatever revision it is.

> Why are there suddenly “b” and “c” variants in ‘v5’ but no “a”?

v5 is v5a. I differentiate minor revisions from major ones, so that reviewers know to expect unimportant changes there. That focuses attention on the major revisions.

> Which version was or is now in the PR branch?

The latest, of course. In that case, v5c.

> In terms of content, there is free text under the versions, which is good, but then there is a privately defined git command (“git rd”) that has no clear meaning for anyone else and a list of commits where it is also unclear what they should be used for.

When I contribute to other projects, I use regular `git range-diff ...`. But in this project, people know what me and what it means. After all, the names I give to branches and remotes locally are not useful to reviewers. What counts is the commit hashes, which are still visible, and of course the diffs.

> You might be better off sticking to the standard of the respective project instead of your personal one.