[mvdtnz]

I am not exaggerating when I say that if I had the power to do so, if you gave me a 10,000 line PR I would fire you on the spot. As far as I'm concerned that is professional negligence.

(Obviously I make exceptions for things like large scale code generation or automated refractors).

[CuriousCosmic]

Yeah I'd agree if there hasn't been prior review and it's not new code. I normally would set my hard limit around 1-2k LOC and I'm expecting quite a few commits to break that up into something digestible or else it's an instant reject.

But for new code? I'm okay with more LOC than normal but I think 10k lines is my hard limit (including tests and documentation). And I expect no more than 500 lines in any given commit/patch. And realistically I am expecting that patchset/PR to be around 50-100 discrete commits/patches in length. Also it's likely that once things start stabilising that patchset can be broken up and merged in chunk by chunk.

And with prior reviews (i.e. merging an incubation branch into mainline) I expect there to be effectively no changes other than merge resolutions to aggregate PRs/patchsets into the incubation branch that is getting pulled into mainline.