[teampractices] Code review social norms

[Subramanya Sastry]

On 03/16/2016 01:33 PM, Kevin Smith wrote:
> I'm curious what the inherent benefit is of having multiple people 
> collaborate on a patch, as opposed to having a series of patches by 
> different people, each of which advances the product incrementally. At 
> least, it sounded like you (Rob) were advocating collaboration for its 
> own sake.

Here are some numbers from the Parsoid codebase to show patches that had 
an explicit co-author tag. This doesn't account for other patches where 
someone other than the author tweaked the patch slightly (that doesn't 
quite count as co-authoring) and merged them.

[subbu at earth bin] git log --oneline --no-merges | wc
    5465   50651  349772
[subbu at earth bin] git log | grep -i Co-author | wc
      18      82    1013

Not a whole lot, but a non-trivial number of patches.

Here are two patches that had the co-author tag:
* https://gerrit.wikimedia.org/r/#/c/238957/
* https://gerrit.wikimedia.org/r/#/c/181177/

Both were tricky pieces of code to get right. Sometimes, it is hard to 
communicate a fix / problem in a patch via review comments. It is easier 
to write the code and show. Sometimes, that is simply the best way to 
address a problem. Sometimes, the original author is lost in some other 
project / is on vacation / ... whatever ... and patch needs to be pushed 
forward.

Subbu.