Slashdot Mirror
Ask Slashdot: How To React To Coworker Who Says My Code Is Bad?
A week ago, you read the other side of the same question. Now, an anonymous reader writes "I have been with my company for 10+ years and have seen many development cycles on our projects. We have a developer intern who has not been on the team for very long. On day one he started ripping into my code on how terrible it is. We have a code base of roughly 50,000 lines of code. When he comes to me with a complaint about the code it is simply because he does not have the experience with it to actually understand what the code is doing. He is a smart guy with lots of promise, he is asking good questions, but how do I get him to look past his own self perceived greatness enough to slow down and learn what we are doing and how we have pulled it off?"
After all, he's fresh from a CS program where they taught him everything.
What political party do you join when you don't like Bible-thumpers *or* hippies?
This is an ancient problem, with 10 years experience I'm amazed you haven't run into this constantly throughout your entire career. New guys (even old guys) perceive everything they didn't write as shit.
How you deal with it is dependent on a lot of things.
First: is he right? Maybe your code does suck. Hell maybe you suck! At minimum. code that has been around for a while, has been written by multiple people over a long period of time, been adjusted and re-worked to meet changing requirements, and been done under a deadline usually does suck at least a little. Admitting this is hard.
New guys want to re-write everything and don't understand the value of code maturity... most of the time a re-write isn't practical, and even the shittiest code usually attains remarkable stability by virtue of having all the bugs pounded out through years of use. Reminding him that this isn't a university project and a certain level of ugliness should be expected might help.
If you don't think he's right, learn how to properly describe why you do things the way you did, and conversely expect him to explain why they are wrong. This is the biggest thing to learn when doing code reviews, and applies here. If you can't objectively describe what is wrong using with references to either standard or internal best practices or conventions, arguing code ugliness just becomes subjective. If you want to defend your code, learn how to describe how it doesn't suck.
Having some company guidelines will really help, because it gives you something to point at in defending a decision. Ultimately what one guy considers good code may be considered bad by another. You are always going to have cases where someone thinks your code is too abstract, or not abstract enough, or sacrifices too much performance for maintainability, or too much maintainability for performance. At least with standards, the new devs will rail against the standards rather than personally attack you, and a standards document is a lot easier to defend (and yet still allows good changes without too much politics of hurt feelings).
Critique is only as good as the suggestions for improvement. So, that's your answer. I feel that if someone has issues with my code, then show me better and prove me it is better. In the end, clarity, code reuse, design patterns, performance, all of these things come to play.
Take a step back and seriously consider his criticism, as if he were one of your 10+ year coworkers. Whether or not he's right informs the right reaction.
You Are Not Your Code: http://sstephenson.us/posts/you-are-not-your-code
Might as well close the forum down, this is gonna be the best answer concerning this issue. if only I had mod points
have you seen my sig? there are many others like it but none that are the same
Recently went through this myself. Despite having used a kanban board, used version control, commented code, written unit tests, etc, some junior devs thought the code sucked. My takeaway was that there were still too many barriers to entry. Too many passwords, not enough installation instructions, etc.
Somewhere in the process of learning to write production ready code, it occurred to me that it was necessary to work the process backwards. Begin a project by setting up your hosting or distribution environment before starting to code. Write unit tests before starting to code. And so forth.
Getting other people to contribute to your project requires the same kind of thinking. Set up a project page before you start to code. Write a vision statement before you start to code. Write installation instructions, coding style guidelines, and operations support instructions before you start to code. That way, as you proceed in the project, you have clearly build up the documentation that other people are going to need to join your project. These things shouldn't be started after the fact.
If you can't point a new dev to a website and say 'download the source and instructions' here, it's probably too complicated and will meet resistance.
From your description, the guy isn't mean spirited. He's likely never had to deal with multiple revision code bases before.
On the other hand, if this is code that has been through multiple revisions and re-purposes, admit to yourself -- it probably is bad. I'm the lead engineer and dir. engineering at a company I've been at for 10+ years, and I'll be the first to tell you that the code-base I am most proud of (30-50k lines of embedded/DSP code, mostly mine) is TERRIBLE! I wouldn't wish that code-base on my worst enemy. But its also been bread and butter for the company for the last decade and is stretched to its limit.
We've had at least two hotshots come onto the project in that time who have been terrified seeing that code-base and declaimed it as schizophrenic at best, and they are right. It is bad code, poor coding practices, and everything else bread out of necessity to keep the project(s) going and alive.
Your mission -- accept that whatever reasons are out there for the code being the way that it is, it probably is poorly structured and could use a rewrite. SO - this is a good chance for him to learn that not every bit of code that should be rewritten can be. Its called business reasons and experience. Whatever the reasons, you probably, as a company, don't have the time or resources to rewrite from scratch (we surely don't!), but a fresh out of school developer probably has not experienced these -non-engineering- reasons for bad code, and certainly was not there for the blood, sweat, and tears that went into them. He won't know about those all nighters that "saved the company" that you and the rest of the team probably went through en-route to this codebase.
I'm sure if he re-reads your internal design specifications, coding standards, and comments in the code he will understand your design.
yep.
What next? "I am some code. How do I tell my new maintainers they suck?"
Signature v3.0, now with 42% less memory usage.
This questioner says he's been at the company 10 years and the new kid is hassling him. That prior question says the guy he's hassling has been at the company longer than the hassler has been alive. If they've hired a 9 year old as a coder then they deserve all the atttitude they get.
As an engineer, I've adopted the maxim that there is no good and bad, only fitness for a particular purpose. I prefer a discussion of trade-offs to statements of principle.
I tend to ask "what requirements does this code fail to meet?" And very often, the reviewer has invented his own new requirement! Depending on your process, your response might be anything from "good point, let's add a test case for that" to "you should submit a Requirements Change Form for that. Make sure to get all the required signatures."
And if the criticism is for something immeasurable like "readability" or "maintainability" you can let your critic make the case to the boss why fixing this code is worth the cost.
[Sir Garlon] is the marvellest knight that is now living, for he destroyeth many good knights, for he goeth invisible.
This, a thousand times this -- but remember, it works both ways.
The professor I worked with closely in university was a strong proponent of this. Sit down, go over the code, look at what can be made better and why, and what may need to stay crufty because it's arcane, and if someone doesn't understand something that's an excellent time to explain it. Walk through and review everything, and don't let your vanity get in the way of writing better code.
He called it egoless programming back in the 90's (no idea how old your book is), and I've found it extremely handy -- objectively look at your stuff and see if he's right. If he's wrong, explain it to him. In the end, the better solution should win out (unless someone is being stubborn or the 'better' solution is technically feasible, which does happen).
Get over the whole "I'm a code guru who knows everything", and be prepared to explain, justify, or even accept that you're wrong. In the end, you review your code to make sure it is as good as you can make it (within reason) and your junior guys get up to speed and understand things better. You get better code out of it, your junior guys get up to speed faster, and you might even learn something along the way.
It's still how I work with co-workers, and it helps in a lot of ways -- instead of saying "here's the solution", you start with "here's a solution, what have I missed?"
You won't agree about everything, but you can pick your battles and try to reach quorum -- the cats don't need to be herded when they're on-board in the first place.
I've seen really awful, and really good ideas from junior coders -- part of the process is helping them understand which is which. In a year or so when he's got more real experience, and knows he can have the discussion, both you and he might actually be better coders for it.
Lost at C:>. Found at C.
"You dare insult my code!? I'll kill you where you stand!"
I've said this at several workplaces:
What the Business values:
1. Correctness
2. Reliability
3. Maintainability
4. Speed
5. Coolness
What the Developer values:
1. Coolness
2. Speed
3. Correctness
4. Maintainability
5. Reliability
Don't believe me? Look at practically ANY open source project. Most are unreliable and impossible to contribute to. They are often incorrect, but they are usually fast and the programmer always thinks what he's doing is mega-cool.
Developers need to adjust their mindset to be valuable to the business, but sadly most business code looks more like the second list than the first.
Peter predicted that you would "deliberately forget" creation 2000 years ago...
We always see bad code as technical debt and treat it as debt. When we have cycles we pay it off, when we don't we get some more.
Having been doing this for 12+ years, I have come to realise I have never liked anyone's code and figured out its more of a style or personality than bad or good. Yes there is bad code, but there is no good code per say. Or put another way, if done right its good and you and I still may not like it.
Its just the nature of the beast, its kind of like driving a car, all other drivers seem bad.