Hacker News Reader

Comment by jchw | Hacker News Reader

IMO, it’s a waste of time. How does the person who wrote the code know what exact thing you don’t understand? You also aren’t asking a question so it’s not really easy to answer. This stuff tends to take multiple review cycles to resolve where a more specific question might actually clarify things in one cycle (It’s never good to assume your lack of understanding means it doesn’t make sense, of course.)

View on HN

For me (and I usually say this instead) it means "This should be explained/documented better for future readability"

> How does the person who wrote the code know what exact thing you don’t understand?

The suggested alternative, "If this is an optimization, can you please add comments?", is a guess. Are we supposed to guess what uncommented code does, and put those guesses in the code review? That sounds even more unproductive than just saying "I don't understand".

It's usually not a waste of time, and the person can expand what the commented line does. A more specific question can be preferred sometimes, like my comment stated.

My general approach to something like that is to ask (in person usually) to sit down with the author for a couple of minutes to discuss it. Getting into a back and forth conversion in review comments is in my experience too slow and cumbersome.

Yes, it's the equivalent of "It doesn't work."

That's not a problem, that's a complaint.

If you don't understand something, explain what you think it should do and where you're understanding is failing.

Depends on your review process - if that comment acts as a trigger/marker for a face-to-face conversation then it makes sense.

If you want to trigger a conversation you could just as easily say:

"lets discuss this"

And avoid putting the burden of ambiguity on the author of the original code under review.

Reviews at Google very often span timezones. Understandably, this sort of approach isn’t always ideal for optimal usage of time.

Reply on news.ycombinator.com