Zulip Chat Archive

Stream: general

Topic: code review effects

Joachim Breitner (Mar 05 2022 at 12:58):

I wonder if mathlib PR could be statistically analyzed to see if https://mobile.twitter.com/iamdevloper/status/397664295875805184 applies here as well :-)

Joachim Breitner (Mar 05 2022 at 12:59):

(and I wonder if this effect has a well-known “Someone's law” name)

Stuart Presnell (Mar 05 2022 at 14:51):

Not quite the same, but this is similar to Parkinson’s Law of Triviality

https://en.m.wikipedia.org/wiki/Law_of_triviality

Stuart Presnell (Mar 05 2022 at 14:52):

“Parkinson provides the example of a fictional committee whose job was to approve the plans for a nuclear power plant spending the majority of its time on discussions about relatively minor but easy-to-grasp issues, such as what materials to use for the staff bicycle shed”

Kyle Miller (Mar 05 2022 at 17:42):

More accessible PRs tend to get a good number of comments (maybe that's good because these PRs might affect more people?) Maintainers tend to ask contributors to split up large PRs, but sometimes large PRs are written competently enough that there's not much to say except maybe naming conventions. This will skew the numbers, since large PRs that need more review will be split up into smaller ones that get more feedback.

While the story going along with Parkinson's Law of Triviality seems on its surface a good one, I'm not sure a committee approving a power plant really should be evaluating the internal design -- that's why they hired the licensed nuclear engineers, right? The bike shed, on the other hand, is something employees will be coming into contact with every day.

Jireh Loreaux (Mar 05 2022 at 20:26):

<insert obligatory reference to Feynman at Oakville here> (technically not a nuclear power plant, but still)

Jason Rute (Mar 05 2022 at 21:47):

In software engineering, we usually just call that “bike-shedding”, if one is spending too much effort critiquing small details of code or a project and missing the more important big picture.

Scott Morrison (Mar 06 2022 at 05:33):

There's also an effect that some subset of the large LOC PRs are refactors. When these are submitted by very experienced people, I think "looks fine" is an excellent outcome (particularly for the large refactor PRs that will quickly rot).

Sebastien Gouezel (Mar 07 2022 at 14:22):

We will have an interesting data point with #12492 :-)

Johan Commelin (Mar 07 2022 at 14:23):

Is this a challenge to merge the PR within 5 minutes?

Sebastien Gouezel (Mar 07 2022 at 14:24):

No!

Last updated: Dec 20 2023 at 11:08 UTC