In my opinion the argument is more "famous paper X by scientist Y was wrong, let's stop citing it". Except that Clean Code isn't science and doesn't pretend to be.
'It may not be possible for us to ever reach empirical definitions of "good code" or "clean code"'
It might seem far fetched that someone might question the benefits of writing high quality code (readable, composable, maintainable, succinct, efficient...) but I've been in this industry long enough (and worked for enough different kinds of companies) to realize that there is an actual agenda to push the industry in that direction.
Some people in the corporate sphere really believe that the best way to implement software is to brute force it by throwing thousands of engineers at a giant ball of spaghetti code then writing an even more gargantuan spaghetti ball of tests to ensure that the monstrosity actually works.
I see it as an immoral waste of human potential.
I read it as an attack against the idea that there are hard and fast, objective and empirically verifiable rules for good quality code. The Clean Code book itself is a perfect example of how subjective such rules are. I feel really sorry for you if the only methods for software development that you know are "brute force it by throwing thousands of engineers at a giant ball of spaghetti code" and sticking a book that has some fanatic supporters. "Readable", "maintainable", "succint" or "efficient" don't really describe Martin's examples and many functional programming enthusiasts would question "composable" too. Yes, I wasted several hours of my life reading that book and I'm never getting them back.
I never said that this is what I believe. I said it's what a lot of people in the corporate sphere believe. It's the opposite of what I believe.
OOP solves complex problems in a simple way.
Functional Programming solves simple problems in a complex way.
Some concepts from FP are useful when applied within OOP, but pure FP is simply not practical. It doesn't scale in terms of code size, it's inefficient, it's inflexible, it takes longer to develop and maintain, it's less readable because it encourages devs to write long chains of logic spread out across many files. FP's lack of emphasis on important concepts such as blackboxing, high cohesion and loose coupling encourages developers to produce poor abstractions whose names sound highly technical but whose responsibilities are vague and impossible to explain without a long list of contrived statements which have little in common with one another.
Abstractions in FP tend to be all over the place. It seems to encourage vague, unstructured thinking. Decoupling state from logic makes it impossible to produce abstractions which are high cohesion and loosely coupled. It forces every component to mind every other component's business.
This is madness. If you don't care about structure, why not just write the entire system as a single file and define thousands of functions which call each other all over the place? You would get the same spaghetti, but you would save yourself the effort of having to jump around all these files which don't add any meaningful structure anyway.
You shouldn't think of it like "I've been fooled and wasted all this time on FP". You should think about it like "I gave FP a thorough analysis over several years and it was a worthwhile experiment which didn't work out but I learned a lot from it".
I also spent quite a lot of time working with and reading FP code over the last 15 years. That's why I can criticize it with confidence today. It was not a waste.
If there's something we maybe agree on, it's that enforcing pure functionality (I guess that's what you're writing about. If you think functions as first class objects is bad, you're incorrigibly wrong) is about as bad as maximal OOP that Martin is preaching. There's a lot of space between these two extremes if you're willing to look. And a lot of money to be made by sticking to an ideology and preaching it.
> 'It may not be possible for us to ever reach empirical definitions of "good code" or "clean code"
If you re-read that line you quoted you may find that it talks of the (im)possibility of defining what "good code" or "clean code" is, not of actually writing it.
A bit like pornography in that way.
I cannot believe what I am reading here.
My open source community knows exactly what good code looks like and we've delivered great products in very short timeframes repeatedly and often beating our own expectations.
These kinds of articles make me feel like I must have discovered something revolutionary... But in reality I'm just following some very simple principles which were invented by other people several decades ago.
Too many coders these days have been misled into all sorts of goofy trends. Most coders don't know how to code. The vast majority of the people who claim to be experts and who write books about it don't know what they're talking about. That's the real problem. The industry has been hijacked by people who simply aren't wise or clever enough to be sharing any kind of complex knowledge. There absolutely is such a thing as good code.
I'm tired of hearing developers who have never read a single word of Alan Kay (the father of OOP) tell everyone else how bad OOP is and why FP is the answer. It's like watching someone drive a nail straight into their own hand and then complain to everyone that hammer and nails are not the right tool for attaching two pieces of wood together... That instead, the answer is clearly to tie them together with a small piece of string because nobody can get hurt that way.
Just read the manual written by the inventor of the tool.
Alan Kay said "The Big Idea is Messaging"... Yet almost none of the OOP code I read designs their components in such a way that they're "communicating" together... Instead, all the components try to use methods to micromanage each other's internal state... Passing around ridiculously complex instances to each other (clearly a whole object instance is not a message).