I am not sure that it worked that way. There were various repositories used for controlling quality. Nobody committed directly to golden; it was gated by builds. Feature and patch branching is a good quality control strategy with pull requests; it is a proven model.
Wade On Sep 11, 2017 12:12, "Matthias Bläsing" <mblaes...@doppel-helix.eu> wrote: > Hey, > > Am Montag, den 11.09.2017, 18:00 +0200 schrieb Daniel Gruno: > > Another approach is Commit-Then-Review (CTR), but that should usually > > be confined to specific features like documentation or experimental > > features. in CTR, you commit directly to a main branch, and if > > someone has a problem with it, they will raise an objection. > > CTR was the modus of operation for development of netbeans prior to > apache and worked good. There was a limit and that were API changes, > meaning all changes that are visible outside a module. For these API > review was required. > > So changes touching API or other global implications could go through > the review-than-commit process, while local changes run through the > commit-than-review (or do nothing) process. > > Greetings > > Matthias >
