- cross-posted to:
- [email protected]
- cross-posted to:
- [email protected]
Image Transcription: Twitter
Giray Özil, @girayozil
Ask a programmer to review 10 lines of code, he’ll find 10 issues. Ask him to do 500 lines and he’ll say it looks good.
I am a human who transcribes posts to improve accessibility on Lemmy. Transcriptions help people who use screen readers or other assistive technology to use the site. For more information, see here.
Good bot.
It’s a human transcriber
Good bot
It’s a human commenter
Good cyborg
Good human
LGTM!
Let’s Gamble, Try Merging!
Why. Whyyyyyy people need to comment this always? Why isn’t just the Approve button enough? I so much hate it.
Ah, that’s too boring. I have a range of responses to pick from to keep things interesting:
- LGTM
- Nice
- Looks good
- Thanks
- Looks great
- :thumbsup:
- Looks good to me
- :shipit:
For me, no text means “I haven’t really reviewed this properly so don’t want to write anything that could be used against me if (when?) this breaks something in prod”
If you’re in a place with codebase analytics you want to have at least one comment on every MR - otherwise the system will start to think you’re falling behind… I hate codebase analytics.
Analytics software like that has made my professional life so annoying at many times.
deleted by creator
In my experience, the managers get confused when issues/PRs are closed without any comment.
Useless comments beat having them pop into your slack to ask “hey, did you review this?” with a link to an approved PR.
True, but the 10 line change can me merged two weeks from now and it won’t make any difference. If you let the 500 line merge languish for two days you’ll have screwed up the work of three other programmers.
That’s why PR should be small. It’s much better to have multiple PRs than a single big one.
Totally fair to have gigantic PR full of boilerplate code, but generally you can split the boilerplate and your feature in 2 PRs, where only the feature will get a proper review.
All of this obviously depends on the criticality of the system :p
That can lead to another problem though, which is that if a developer knows a merge is only part of the whole change, it becomes easy to assume any issues will be handled elsewhere.
How do you improve on this?
1 bigger PRs, but with multiple smaller commits, so reviews can review by commits?
Or split them per commit at the minimum if you don’t want to create a separate PR.
Y’all need to get yourselves some PR review automation in place. Stop wasting time on trivial reviews and requesting changes for common problems so that when you ping a colleague for a code review, they know it’s important rather than a simple request for a thumbs up.
This can be called the bike shed effect or Law of Triviality. It’s not just programming where a simple digestible idea will be contested because it’s easy to poke holes, while something more complicated and more consequential is much harder, so it gets little to no resistance.
Really grinds my gears to get a review request for a huge PR with almost no context and no instructions on how to run the code / test data.
I’ve even had one guy ask for a review who hadn’t even manually tested his own codes happy path. Sure he wrote some unit tests and those ran but once you actually tried using the code in the app all kinds of exceptions and weird situations came up. No idea how people dare do that shit.
Poor management and time pressure are the answers. It often looks better to push some shit code and then push the fixes later than it does to take twice as long to verify a good change.
No idea how people dare do that shit
perhaps we should introduce the PR Wall of Shame for exactly those situations. I mean, obviously you don’t want to strangle productivity by naming and shaming people for every single small mistake, but such egregious violations like not even click testing the happy path should be used as an example of what fucking up looks like
Playing the blame game, especially publicly, is a bad way to encourage psychological safety in a community. See Linus Torvalds for examples (he has recently become a bit softer with his feedback).
A better option is to make clear expectations of what a good PR looks like. Then if expectations are not met, you can give 1 on 1 feedback. Don’t just blast a noob in public or you can leave emotional scars.
Don’t just blast a noob in public
true, but also when someone who’s been around for a while and ought to know better, it can really help to remind them that they’re not above the rules
I know this implies that the reviewer didn’t care to read the bigger PR, but I think this might actually be legit. If your PR is only 10 lines long, then chances are those lines are very dense, or intricate in some other way. However, if you submit 500 lines, then it’s probably mostly boilerplate code with trivial adaptions.
No-one in their right mind would submit 500 lines of substantial code.
Oh sweet summer child…
Has no one here ever worked on a new project or even a new feature in a decently sized codebase? Working exclusively in maintenance / minor change mode has to be exhausting.
Depends on what you classify as “minor change”. When I took up my first professional project, I found a plethora of little things to improve which would make users happy. That was very satisfying.
On the other hand, writing yet another module that displays a list of Foos, lets the user create a Foo, show the details of Foo, update it, and delete a Foo, becomes dull quickly, despite being a “new feature”.
500 line of substantial code happens once in a while in my team. Goes beyond that sometimes.
If a programmer does 500 lines, how much work will they get done before their heart explodes?
@[email protected] is right. What the lines are is very important. But it also matters over what time period.
If we do say, 20 lines per workday for a month, your heart will be fine!
Your nose might fall off though.
LGTM ✅
A couple weeks ago I was tasked to review my own code. It was over 10 lines, so I just passed it.