pull-request tips

README.md

@@ -24,3 +24,31 @@ And update like so:
     raco pkg update --update-deps pollen
     
 Official mailing list: http://groups.google.com/forum/#!forum/pollenpub
+
+## Pull-request tips
+
+I welcome pull requests. But accepting a PR obligates me to maintain that code for the life of Pollen. So if I seem picky about which PRs I accept — yes, because I have to be. No hard feelings.
+
+* There’s plenty of room for improvement in the Pollen code, because every line of it has been written against the backdrop of ignorance and fallibility, mostly my own.
+
+* I don’t necessarily prefer PRs to issues or feature requests. A good description of the problem with a working example is better than a half-baked PR. I can often fix it in less time than it would take to review the PR.
+
+* If you want feedback on a potential PR, I recommend posting to the [Pollen mailing list](http://groups.google.com/forum/#!forum/pollenpub) rather than here. Because more people will see it.
+
+* Small PRs are easier to accept than large ones. Large PRs should have a benefit worthy of their complexity.
+
+* I consider every PR, but I can’t promise detailed code reviews or comments. Helpful Racketeers can be found on the [Pollen mailing list](http://groups.google.com/forum/#!forum/pollenpub), the [Racket mailing list](https://lists.racket-lang.org/), and the Racket [Slack channel](https://racket.slack.com/).
+
+* PRs should be necessary, in the sense that the proposed change can only be accomplished by patching this repo. (Corollary:  features that can live in a separate [package](https://pkgs.racket-lang.org/) probably should.)
+
+* PRs should avoid introducing magic behavior (aka the [principle of least astonishment](http://wiki.c2.com/?PrincipleOfLeastAstonishment)).
+
+* PRs should forbid as little as possible. In particular, PRs should avoid enshrining personal preference as default behavior (because others will have different preferences).
+
+* PRs should avoid reinventing features that already exist in Racket.
+
+* If you’re new to Pollen or Racket, your PR is more likely to be declined, because certain things you perceive as bugs are actually features, certain things you perceive as missing are actually present, and certain limitations you perceive as surmountable are actually not. (See also point #1 re: backdrop of ignorance.)
+
+* PRs that could have unit tests, and don’t, will be treated harshly. As they should.
+
+* PRs that want to amend Pollen’s public interface receive the highest scrutiny.