3 ms·
Elements of this plan sound good. Obviously musk doesn't have time to review it all... But saying 'print out a couple of pages of your recent code you are pr
by londons_explore 4y ago
Elements of this plan sound good.
Obviously musk doesn't have time to review it all... But saying 'print out a couple of pages of your recent code you are proud of and bring it to a review with your tech lead' is a good way to discuss coding style.
The 'you are proud of' is key to this working though - all your code is visible in the VCS anyway, but asking you to pick a good bit to discuss seems valuable.
- fzeroracer 4y agoIf a tech lead told me to print out some code and bring it to a review, I'd strongly reconsider the team I'm on. Not only is it among the worst ways to review code, but it also includes a lot of irrelevant information.
- systemvoltage 4y agoWhy? Printed code is code, NASA/Military does it for formal review. I've worked on avionics systems that are developed in similar fashion. You ought to be able to explain what it does. Most of it is going to be plumbing. Call this, fetch that, join foo, collate bar. Strong arming Elon's intentions, may be he is used to perhaps how code reviews work in Space industry? Sometimes old school ways are fine, under the hood it is just code.
- fzeroracer 4y agoNo, no they do not. I've worked for the government and in that sector. We did not print code to review it. That might've been a thing a decade or two ago, but it is not a thing now and for good reason. And I would strongly question the technical capabilities of a team that was not able to adapt to modern tools even for older codebases. What exactly is your citation that they still do that? It is simply much quicker and much more efficient to do something like a proper code share in an office with one person driving the code and being able to jump between areas of the code. To say it bluntly: Anyone that uses printed code for code reviews is wrong. It assumes all functionality is colocated and nearby. It wastes time by forcing people to index through multiple pages of paper. Anyone that demands it should see prompt pushback.
- systemvoltage 4y agoI worked in 2010, so yea it might have moved on. It's probably in powerpoints now ;-) I worked in military (not space) though. I still know people there and I'd guess it is not that modern. We'd also make bluebooks in 2010 with code printed and neatly bound in them.
- throwawayx130 4y agoI worked both in the military and the military industry in the early 2000 and we never printed out any code for review. We did have very thorough design documents and manuals and all of them have been printed (including documented function headers) as the contract dictated, but we only used the original word documents and the version control system to review our code.
- ergocoder 4y agoPeople keep saying printing code is bad for twitter engs. If I'm forced to print code, I would probably do it. It is essentially a few clicks taking a few minutes of my time. On the other hand, your tech lead reviewing code on paper is going to be extremely painful. The tech lead who is forced to review code on paper would probably quit first.
- minraws 4y agoFrom this perspective this seems very reasonable, if I picked some code which I like it will be very hard but I believe it will force me to consider what I don't consider in the day to day. Am I really proud of what I do? And what do I do really? Maybe with a bit of back and forth with team mates it might help people like me as well who struggle with imposter syndrome on a daily basis feel like they can be proud of things they do.
- kotlin2 4y agoJudging the sizes of the stacked pages from the last 60 days of code would probably be a pretty good indication of productivity. Perfect? No. Good enough? Maybe.
- code_duck 4y agoAs in, the fewer pages, the more concise they are? What about clarity, and which languages are we talking about? Bonus points for Java? I think it’s been fairly well determined that LoC isn’t a good metric for judging productivity.
- kotlin2 4y agoI think LoC has been determined to be a bad productivity metric because it’s gameable and incentivizes bad behavior. However, that doesn’t apply here as the engineers didn’t know they were going to be evaluated by LoC a priori. I struggle to think of cases were lines-changed wouldn’t be correlated with the productivity of engineers working in a consistent environment if they weren’t trying to game the system.
- code_duck 4y agoThe problem is some languages are more verbose than others. Considering boilerplate (which is often automated these days) it extra has nothing to do with productivity. More lines doesn't mean more action - I mean, we're on a Lisp-derivative forum. And sure, if I was getting judged by LoC I'd just paste in a library versus including it, which would generally be worse.
- whimsicalism 4y agoYeah, no.
- esoterica 4y agoI never thought I would see someone on HN unironically advocate for counting LoC to measure productivity. Unbelievable how this guy's cult of personality poisons brains.
- vkou 4y ago> is a good way to discuss coding style. Is scheduling every single person in the company to have a discussion about... coding style solve any particular problem Twitter has? This reeks of 'I am the new owner, I need to do something, this is something, everyone will now be doing it.'
- yetihehe 4y agoIt helps to weed out total slackers, of which there could be non-negligible percent. If you have any shred of decent code - you are ok. If you can't produce a page of decent code after working a year - why would anyone keep such programmer.
- scrubs 4y agoAnd why weren't they already canned? Slacker tls and managers. Next thing you know Elon is looking at himself. Yeah bring code so we can talk about it is fine at the bottom level. Dumb for owner in first week. The good programmers know nothing will ultimately come of this codiscovery process thus making a bad impression on the good employees
- batmanturkey 4y ago
- aniforprez 4y agoWhy in the Sam Hell would you need printouts. Literally look at git histories and contributions per user. This is bending over backwards to explain what is most likely a simple power play that most probably backfired
- yetihehe 4y agoSo that they can show code they are proud of, not just average code or code written under pressure or when they had to do quick fixes.
- 4y ago
- kitsune_ 4y agoWhat you say is of dubious value (or even negative value). I imagine Twitter has code reviews like most other places anyways. And even if were not of zero value, it's not what they were told to do anyways. Printing out your code of the past 30 days is a demeaning power play to test your loyalty. They could just look at your pull requests or commits themselves.
- sebastialonso 4y agoWhat possible value to the company can come from the CEO doing this stunt for coding style? Don't you think there are already company-wide guides for this? It sounds like something that someone from outside the software engineer world would think it's valuable.
- 8n4vidtmkvmk 4y agowhat even coding style. we have so many lints in place that almost all code is homogeneous.