Hacker Newsnew | past | comments | ask | show | jobs | submitlogin

What a weird thing to say. I can evaluate any code. If it’s code I’m unfamiliar with it takes longer, and the result is more variable, but it’s far from impossible.

It’s certainly more reliable than the evaluation of the person whose job depends on it being evaluated as correct.



Given 24-48 hours in a gigantic tech company’s code base made up of dozens of teams and hundreds (thousands?) of logical services? No, I couldn’t tell you the “quality” of the code.

Someone who doesn’t understand engineering at that scale would presume it’s the structure of the if statements. And sure, if that’s what you’re looking for, print out some lines and let a person lacking all context take a look.


> Given 24-48 hours in a gigantic tech company’s code base made up of dozens of teams and hundreds (thousands?) of logical services? No, I couldn’t tell you the “quality” of the code.

You could probably call out the bullshitters, though. They aren't refactoring the code base.


I would guess anyone calling someone out 48 hours into a huge codebase is themselves the bullshitter. Maybe if that 48 hours is work time and presentations/walkthroughs and the person catching up is a seasoned director (they tend to have to do this quick ramp of huge projects more often than an engineer). But no one is going to look at Twitter's code base and make any meaningful conclusions in 48 hours.


Agreed. I had the opportunity to audit a FAANG codebase (many many repositories) a while back and while I'm a lowly engineer, I was able to start understanding quite a bit after a few weeks. Obviously there's no way I could contribute or make recommendations in that short time, but obvious smells start working their way out after exposure. The clarity only increases with the amount of time spent immersed in it.


Off course you can have an opinion about any code. It will just not always be correct.


I'd like your opinion on the MGLRU code they just merged in the kernel. It's a self contained piece of code, you can look at it and print it on paper.

Or the folio patchset. People had a variety of reactions to it, whole lot of energy was spent evaluating that code (which in the end also made it in)

The person whose job it was to make the patch and who has interest in it being correct is not necessarily less reliable than a random HN commenter

You would probably think otherwise if it was your code and some other random commenter with no domain knowledge evaluating it, wouldn't you?


> If it’s code I’m unfamiliar with it takes longer, and the result is more variable, but it’s far from impossible.

Yeah, I'm sure it's of value to have Tesla engineers come in and spend the 3 months to a year getting enough context with the code base to be be able to produce useful and informed critiques.


[flagged]


You don't have to evaluate other peoples code or something?




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: