> I’ve recently come across a discussion where a new developer joined a team and faced over 300 PR comments on their first contribution. Most of it was stylistic nitpicking. This isn’t just unproductive, it’s outright toxic.

For me this says more about the company culture than any inherent flaws with the code review process.

My team has a bike shedding sort of problem where a 100 loc PR will sometimes get scrutinized to hell, but a 3,000 loc PR will get LGTM'd by enough of the team to be merged before anyone that actually cares gets a chance to look at it. I would say the second half of that is the much bigger problem.

People know who to ask to get a quick lgtm.

I don't know what to do about it. I can't make people actually review. I've had unrealistic dreams like holding people responsible for things they approved that had bugs, but anything like that would be supremely unpopular. I could do something like requiring review from the team members that actually review, but they already feel overwhelmed by being the only ones that actually review.

We tried setting soft limits on the size of PRs, but that comes with a lot of PR that are hard to review because the work is poorly divided and doesn't make sense in isolation.

Whenever I encounter a pull request that I find many issues with, I ask to meet with the engineer and review it one on one. More than half of the workplace problems engineers have is due to their introverted nature and their refusal to get on a teams/zoom call to explain their issues and get resolution. Comments are a really poor mechanism for teaching programming best practices.

Personally I have a strong distaste for projects that try to use some metric for how long a function should be, e.g. line count or cyclomatic complexity. I'm not sure if this one is better automated, human judgement for what makes sense seems better to me. Sometimes the cleanest and highest performance way to write some code is going to be basically one relatively large function. If there's a function complexity/size lint I'll often chunk off helper functions that are virtually useless outside of that specific function, and in doing so, often make it harder to follow what's going on.

This isn't universally true of course, but it's a "when not if" situation if you have a hard lint limit and it's not very high. (e.g. In my opinion if you're going to have a lint for this, set it to something rather obscene, something that would be extremely hard to cross without doing something clearly awful. In my opinion it's always been set at least 4 times lower than it should be.)

One core issue is that PRs are often too large. I don't believe there are 300 comments on a small PR unless the team is dysfunctional.

Aim to make them as atomic as possible. If it involves adding a significant feature that requires substantial changes to other modules, create a separate branch for it. Within this branch, break the work down into a series of smaller PRs.

When PRs are too large, the process usually suffers from insufficient reviews, excessive nitpicking, or (most often) a combination of both.

I can't imagine going back to working on codebases in languages that don't come with strongly opinionated defaults.

I'm very lucky to spend the majority of my dayjob hours working on Rust code, so everyone just runs cargo fmt && cargo clippy, and this in enforced in CI. You can't even publish a PR until those basic bars have been met.

I can't imagine the absolute insanity of working on JS projects where there are more ways to do things than there are people on a team, and where getting those people with strongly held opinions about things that ultimately just don't matter to agree on style and convention is like pulling teeth.

I'm shocked how many companies still don't focus on DevX and wonder why not only is productivity in the gutter, but morale as well. There's so many companies that help solve DevX problems like this, that's it's practically easy to throw money at it.

Another angle is that a PR with over 300 comments is probably way too big.

Many small focused PRs is IMAO much better than big PRs with weeks or months of work in them.

But that requires fast and friendly PR reviews!

Recently I allowed everyone in my team to push to develop without PRs and even without doing feature branches. We review the code all together (max 2 ppl) just before the release and that's it. But great CI pipeline is must for this. It will get even better soon when AI will be able to slightly refactor the code.

Yes!

I’d also add that it’s helpful to have a review guideline.

When I get to the PR phase, I’ve already done the design work, considered the trade-offs, and am asking folks to check that the code is mergeable per our guidelines.

A pet peeve of mine are reviewers with a bone to pick who will leaving blocking comments to redo the work in the way they would have approached it.

A guideline helps here because you can, along with your team, manage expectations for the process.

Some folks like to throw up code for critiques. They use PRs to get feedback on the design/approach itself. That can be useful!

Going in to manage expectations helps smooth things along so that you’re giving the constructive feedback the author is looking for.

This is really important to get right. There's a balance as you want to avoid introducing crap but a little restraint and wider awareness makes a huge difference.

I recall being significantly put off after making my first significant contribution online to a non-trivial repo.

I was a volunteer and had been making various very modest contributions (ie simple fixes) along with efforts around docs and answering questions from users. A major contributor was known for being cantankerous. He knew what I was bringing in general, and seemed grateful yet didn't hold off with pretty forceful feedback on my first attempt at something more substantial - I had taken a day off to get it in good shape, I knew it was solving something with broad use (ie not just for me!). There were a litany of points. I fixed several but each time there would be more, it was like water on Gremlins! I gave up and stood back for a bit. Ironically it would have been trivial for him to take the code in but it sat forever and the whole thing was a waste of time but I guess I learned a few things!

I try to encourage standards but not officiously!

Part of me wonders if the real issue is we don't have an author/editor type system. If the reviewer/editor could just make the nitpicky changes in about the same time it takes to call them out, the relationship might be much more healthy. Things could go back and forth in a much more healthy way.

I have always found articles/discussions of code reviews fascinating. In the 10(ish) years I've been employed as a programmer I've never worked in a department with more than ten people and every time the work is distributed such that each person works as an IC on their own siloed project(s). Receiving anything besides the "LGTM" rubber stamp or nitpick comment seems almost impossible.

Watched a guy get fired mostly because he made code review such a nightmare for everyone else while getting lapped by everyone in terms of actual output. I’m not sure there’s much to be done once you’ve hit “toxic.” The only question is whether the standard can be reset for the whole team.

Non blocking code reviews ftw

https://itnext.io/optimizing-the-software-development-proces... Optimizing the Software development process for continuous integration and flow of work

https://medium.com/itnext/how-feature-branches-and-pull-requ... How feature branches and pull requests work against best practice

I've worked with plenty of devs who embody the toxicity here, lots of stylistic nitpicks and don't see the big picture. All the noise generated distracts from catching the important stuff.

People being an obstacle to positive change, because they believe a better change is possible.. when the changes wouldn't in any way prevent further positive changes from happening.. are the bane of my life. 'bikeshedding' probably covers it, but i feel like if someone could come up with a better metaphor that wasn't specifically about triviality or level of comprehension it could be a profound contribution to human behaviour

I’ve had a PR into an open source Anthropic library open for almost 2 months. It passes all of their rules, fixes an obvious bug that causes functional issues, passes all tests, and has a description that is descriptive with a complete user story explaining the functional issue with images. I really like using Claude and I like Anthropic’s mission but it definitely kills morale to fix something and be put into PR purgatory.

The only thing I disagree with him on this is that function length doesn't matter. Longer functions do tend to be harder to understand. I will admit that sometimes a lot of helper functions can make it harder to piece together, so you can overdo it, but usually putting one or two of them back into the main function gives a very readable function.

So I would tend towards more smaller more decomposed functions with descriptive (but concise as possible) names by default. This is very different from old-fashioned C standards, which I feel are very outdated for modern tooling and hardware.

It's definitely important to have formatting standards to avoid arguments about that. But you are never going to completely eliminate stylistic things because part of that is function decomposition or intersects with the actual substantive design. Because some people feel quite differently about function size for example.

I don't know the original story, but 300 hundred PR comments on a first PR (or any PR for that matter) is insane, and my guess is that somebody on the team is having an argument back-and-forth in that PR, rather than focusing on whether the code should be merged.

If that happens to me, I would kindly invite them to talk about it elsewhere and give me pass, because the fact that you guys are still debating it, means that it's not a decided thing, and you can't prosecute me with a law that hasn't passed yet. Once it's actually settled, I'll be happy to go back and refactor it, but it is illogical to make it a blocker in the present.

It also sounds like something is wrong with the on-boarding process. The first PR should not be of such a nature that would invite loads of comments. It should be something that is functionally simple, and already has an agreed-upon design.

It's never the process it's always the people and the underlying competence gap that kills morale.

Code reviews are weird. I joined a very small team as a new developer. I got a task, solved it but at code review the senior developer told me to solve it in another specific way. I coded it and it was fine. The situation repeated for few next tasks. It was a bit frustrating because I felt like doing the job twice while "pre-review" of my tasks by the ultimate code reviewer before I started doing them could save both my time and his. But it just wasn't part of the process.

Does such thing have a name? A quick check up with a reviewer about whether my plan for implementing the task is align with their preferences?

I found two ideas / techniques helpful in this context:

1) Conventional comments (https://conventionalcomments.org/) as an (agreed-upon) language to be used in PR comments

2) Ship / Show / Ask (https://martinfowler.com/articles/ship-show-ask.html), where "Show" and "Ship" are non-blocking PRs (or even directly committing to trunk, if you use trunk-based development), since not every(!) PR needs reviewing and/or should block the PR creator

> Define (and Stick To) a Style Guide

We have some styling rules that are so much in the "Style Over Substance" category that I have a strong tendency to subconsciously categorize them as "not important, at all". Especially when the IDE doesn't respect them when you use the code generation.

But as it was a strong irritant for some PR reviewers, and as I couldn't force my mind stop doing it, I automated it by writing a custom linter.

GIT seams to be optimized for network of trust. With one person at the top approving what gets merged into the release.

This person of course does not do all of the verification, other then broad strokes of what the change does, and who wrote it, reviewed it and tested it.

I feel like companies do not want a large tree like structure for their development teams.

Without a network of trust it can become mob rule, which is what this article appears to be describing.

I usually adopt a policy with my teams:

You are not allowed to complain about style in code reviews.

If it’s important enough for your to comment about, it’s important enough for you to add a linter to the build and block the build if the style isn’t met.

If it’s not worth enough of your time to add a linter, it’s certainly not worth your peer’s time to deal with comments about style.

Per the article, not the pull requests itself, but the tedious bikeshedding around them. The article's headline is a bit misleading.

If you join a company and are confronted with dozens of PR comments - assuming this process isn't sufficiently well documented - see if a tech lead or manager will host an open forum review. Review the code and the comments and discuss + document. A one off discussion won't be a permanent fix, but it can help!

Where do people find these places? I have a hard time getting management to accept that more restrictions on who can review is not a bad thing.

Nearly 90% of our PR’s have zero comments.

I do have a hard ban on formatting comments though. If it’s not covered by the linter/formatter everything goes.

I recommend using something like https://github.com/erikthedeveloper/code-review-emoji-guide in code reviews to signify what your expectation is exactly.

Combine this with a large timezone difference and you get a (literally) slow motion disaster. Imagine waking up and finding your change thoroughly reviewed but not approved because of a triviality. Not a great day.

200 comments mostly about stylistic nitpicking is actually a language fault, not a team fault.

Languages like Golang are specifically designed to remove any stylistic nitpicking.

Every language must follow. It's suicidal otherwise.

I worked at a startup where a developer used to bully other developers in PRs. Management looked away. I warned them but they didn't listen. That team is now dissolved.

I have quit places because of their PR process.

...just how big was that "first ever contribution" pull request that it gathered 300 comments? My first commit in the current company was a 2-liner addition to an existing function plus another 20 lines for a test that verified that indeed, the change does affect the outcome.

I wholly agree with this, and I would add worthless Sonar opinion reports th the list.

When it comes to code reviews, the return on investment faces the Law Of Diminishing Returns. While many of the comments made in code reviews might be interesting, they are not so interesting that they pay for themselves. If you put some dollar value on the time invested, you'll find that the vast majority of this process is simply burning money. And not only money, but also, as this article says, morale.

A curious fact about the politics of modern tech teams is that some of the same people who consider themselves "anti bureaucracy" are strongly in favor of this type of bureaucracy, even when it brings no measurable benefits.

My opinion of code reviews has become more negative over time, mostly due to the fact that I have not seen it achieve reliable positive outcomes.

I'll give specific examples:

https://www.futurestay.com had the worst code that I'd ever seen. When I joined the company I was horrified at the level of tech debt. And yet, for years, they had a rule that at least two engineers had to review every PR. So this terrible code, the worst I'd seen in my 24 years of coding, had been approved by two engineers.

And likewise:

https://openroadmedia.com also had very bad code, and also had a rule that nothing could be pushed to production until at least 2 engineers had reviewed the PR.

In both companies the code review slowed down the deployments, slowed the down the team, and created a culture of internal sniping, while leading to no improvements.

But how can I say the code was objectively bad? Because many bugs were found in production. And what would reduced the number of bugs in production? More tests, especially end-to-end tests. And so I eventually came to this conclusion: it is best to skip code review, and instead have the team invest that same time in writing more tests, especially high level tests. For the most part, if the engineers on a team are bad then the code reviews will also be bad, but if the engineers on the team are good, the the code reviews will not be needed. And in both cases, the path to improvement comes from writing the kinds of tests that ensure no bugs get into production.

As I've grown in my career, and taken on higher level management jobs, I've also realized that code reviews do not scale to large-scale leadership, but tests do. Code review seemed like a good idea when I was leading a team of 3 engineers, but not when I was leading a team of 30. When I was leading a team of 30, the only tool-of-oversight that worked for me was automated testing. And so I've concluded this is where the effort should be made. Sitting at my own computer, I cannot review the work of 30 engineers, but I can still run the tests on the various software, to see if they are passing. In particular, I can look at API interfaces and then change the dummy test data to break the tests, and this lets me see quickly thorough the test coverage is, and how prepared we are for unexpected shocks.

When I am running a team I will assign software developers the task of writing various tests, including high-level end-to-end stress tests. These tests sometimes reveal problems. We then respond to the problems. But we don't waste time responding to problems that have not been proven by a test, which is to say we do not do code reviews. Code reviews are all about predicting what might be a problem. But much of the concerns are phantoms. It is better to respond to real problems that have been revealed by tests.

I've also come to realize that software developers, quite naturally, develop strong opinions about questions of style, and yet these questions have no long-term impact on the health of the tech in the company. Other than enforcing the rules of some automated linter, the time spent on issues of style are 100% wasted. If you allow the team to spend even one minute discussing issues of style, then you are setting money on fire. But I know this opinion is unpopular, because software developers enjoy enforcing their own opinions about style. But it is all bikeshedding, it has no real-world impact.

I have the impression that there is some middle era, in the career of a software developer, where concerns about issues of style and organization tend to peak. The junior developer does not know enough to care about such issues, but somewhere between 4 years of experience and 10 years of experience, these issues feel important. I think you need more than 10 years of experience to see the waste. In particular, you need to run one large project where the team invests a lot of time into code reviews, and you need to notice how much tech debt builds up, despite the code reviews, to realize that code reviews do not offer a path forward.

On recent projects I've told the team there will be no code reviews, but instead we will focus on building tests. I get a surprising amount of pushback on this. Less experienced software developers get angry with me and tell me that I'm being unprofessional. In some sense they are correct, in the sense that "professional" refers to "standard norms that are accepted by a profession" -- I am deviating from those, clearly.

The most reasonable criticism I get is that code reviews could catch not just problems of style but problems of algorithms. What if, they ask, some junior developer introduces code that is ignorant of the implications of Big O Notation? What if they introduce code that runs in polynomial time? But I would ask, how do we know if it is running slowly? We can only know that through tests. So lets build tests that measure time, and stress test with large loads. Big O Notation offers an excellent example of where software developers can worry about the wrong thing: what if a junior level software developer writes code that runs in polynomial time but on data that will only ever have a few hundred records? While the algorithm might be sloppy, the code will run quickly because there are few records? In that case, any time invested to find a different algorithm will be a poor investment. Once I'm running a team of 30, the only thing I care about is whether code is actually slow, and I discover that through end-to-end stress tests, not code reviews.

Kent Beck, when he invented Xtreme Programming, also popularized the phrase "Do not write a comment, instead, write a method with an easy-to-understand name that communicates what the comment was going to communicate."

I've come to a similar conclusion. Do not write a comment in a code review, instead, write a test that would catch whatever danger you want to warn about. And if you cannot find a way to express your concern as a test, the return-on-investment of worrying about that concern is probably zero, so we should ignore it.

[dead]

[dead]

[flagged]