On Tue, Nov 05, 2019 at 06:37:02PM +0100, Nico Huber wrote:
I mean "rubber-stamping of *huge* commits". That huge that it is obvious that no review happened, e.g. 1k+ LOC copy-pasta. Also, the guidelines say "This means you shouldn't +2 a patch just because you trust the author of a patch [...]" I didn't want to quote the whole paragraph, maybe I should have ;)
Some of the mega patches are copies of a predecessor chip (with the minimum amount of changes to integrate it in the build), that are then modified to fit the new chip.
The idea was precisely to ensure that changes between the generations are visible in the history.
That was considered an acceptable strategy for a long time (after we broke up 82801xx because fixing one variant broke another). We can certainly revisit that, but to me that change appears to have happened gradually and without much notice.
In case of really huge commits, the only way to return to a clean Git history is a revert. So that should be easy to measure. But really, what are we arguing about, I mean cases where a +2 already makes it obvious that the reviewer doesn't care about future maintenance nor a useful history.
Again, obvious to whom?
Next: When is the penalty implemented? If it happens right before my vacation, I might not care too much ;-)
I don't think it matters much when it starts.
Under your scheme, 1kloc added is about 7 weeks. Too long to wait it out, not long enough to be practically permanent (but as you note later, that penalty isn't automatically lifted after that time anyway, changing the mechanics), so timing can make a big difference.
Not sure if this is a misunderstanding. "They can only return...", isn't supposed to mean that this would happen automatically.
That wasn't clear to me, even if that's a possible interpretation.
Regards, Patrick