09-12-2020, 02:24 AM
Looking at the git history, Tiger seems to have made a habit of committing direct to master. His latest PR was merged on the 27th of July (#4780). Yet the git log shows 56 commits by him on master since then:
Some of those commits are quite small. Others have changed several hundred lines of code.
IMO nobody should be committing straight to master. Not just because it sidesteps review, but also because it doesn't generate notifications. If there's a rush to get some fix in, then I would say open a PR and self-merge it with a note saying why it's urgent. I understand that PR review can be frustratingly slow and maybe we could all work to improve that. But I don't think this is the right way to speed up development.
Code:
$ git log --author="Tiger Wang" --format=oneline --since="2020-07-27" | wc -l
56Some of those commits are quite small. Others have changed several hundred lines of code.
IMO nobody should be committing straight to master. Not just because it sidesteps review, but also because it doesn't generate notifications. If there's a rush to get some fix in, then I would say open a PR and self-merge it with a note saying why it's urgent. I understand that PR review can be frustratingly slow and maybe we could all work to improve that. But I don't think this is the right way to speed up development.