Does this hold true for you on smaller projects too? Ruby gems, for instance...?
I personally strongly prefer to see a diff via Github, especially for smaller features on a small library / gem -- although on larger & more popular projects (lots of pull requests) it definitely makes some sense that you'd prefer to have a 'heads up' of sorts via email first to confirm that it's even a desirable feature.
Really just wondering here, I recently started semi-blindly submitting pull requests with feature additions or small bug fixes (with very verbose pull request messages + solid but concise commit messages) to various library-type projects so I'm legitimately curious as far as how far to take this logic...
I think email is a better forum if everyone involved is disciplined about factoring commits well, writing good messages, using an email client that formats properly, and is fluent with the tools. PRs are more intuitive, more tolerant of workflows, and more accessible to newcomers. At this point, nothing matches the archival quality of mailing lists. When PRs are rebased/refactored as a result of reviews, the original versions generally become orphaned, so the discussion vanishes.
When there are multiple independent discussions spawned from different parts of single patch, as is common in less disciplined communities, the granularity of email is too coarse, and threaded line-comments are much better. On the other hand, patches on mailing lists frequently spin off into higher level design/philosophy discussions, which isn't appropriate in line comments.
I personally strongly prefer to see a diff via Github, especially for smaller features on a small library / gem -- although on larger & more popular projects (lots of pull requests) it definitely makes some sense that you'd prefer to have a 'heads up' of sorts via email first to confirm that it's even a desirable feature.
Really just wondering here, I recently started semi-blindly submitting pull requests with feature additions or small bug fixes (with very verbose pull request messages + solid but concise commit messages) to various library-type projects so I'm legitimately curious as far as how far to take this logic...