This might be obvious to some, but I've literally only realized a few months ago that when I write text, I will consult a thesaurus, whereas for some reason I've neglected to do the same when I code. I would spend hours agonizing on how to call a module, wondering if I won't need the same word later for something else, or if it might be making things a bit ambiguous for the next guy.
So I've started also using a thesaurus when I code and so far I'm happy with the results. It's been helpful in accurately framing some of the hazier concepts with interesting names. I also revisit the naming much less than I use to, because I'm comforted with the notion that if I couldn't find a better word with a thesaurus, then maybe there isn't an obvious one. I also now use words that are clear, but that I so far neglected to include in my programming vocabulary, because other more prevalent ones would jump in my mind and put a shadow on my naming creativity.
So, if you need some help naming modules, functions, variables and db entities, a thesaurus could help. Though, take care to not abuse of it.
I do this too. Though my thesaurus doesn't always have the right technical meaning of certain words.
I'd love to have a "programmer's thesaurus" with more detail on names that appear frequently in code. For one of the two hard problems in CS, the art of naming things has little reference material beyond raw source code.
I'm using dict.cc for that purpose, which is not just a thesaurus but also a translation tool. Sometime I find the right words in my native language but not in english. Or, the synonyms are not satisfying, but switching between languages leads to words more loosely connected to my initial word, thus extending the search radius in a useful way.
Naming things to be clear in the beginning, stay clear as the system grows, is quite easily one of the hardest things in developing software.
Over generalized or specific names cause far more confusion and productivity loss than we think.
So take the time to learn what you're working on and how it works at the micro and macro level not only in the context of the application itself, but where and how the application fits in it's ecosystem.
Learn the lingo that your software's ecosystem uses, instead of inventing, by accident or purpose, your own clever and tricky rationalizations ending up in even more confusing concepts and names describing the same things.
The purpose of naming is similar to the purpose of software; creating clarity and empowerment, instead of more confusion and conceptual gymnastics.
For more rigorous naming, Eric Evans (in his excellent book Domain-Driven Design: Tackling Complexity in the Heart of Software) recommends documenting (and maintaining) your product's "Ubiquitous Language".
When developers (and the code), product managers, domain experts, and end users all use different terminology, then cross-discipline communication difficult and imprecise. A documented Ubiquitous Language enforces shared terminology (and meaning) for all communication within the team and in the code.
I am glad someone wrote this article - Now I know I can point it to my colleagues who are often baffled at the amount of time I spend thinking about names :)
I have found that when I can't find a good name for a method or a class, it's because I haven't really thought the reason for the same through. This is especially true for classes where I often realize that I am making the class do too much, or too little.
This book, "Clean Code" by Robert Martin has it's second chapter dedicated to naming things. A whole chapter! The book, alone for that chapter is worth the money. If you haven't read it, I highly recommend it.
I find that thinking about good names for things can help guide you to design better code.
As Lars mentioned—if something is hard to name, it's probably because the thing you're naming isn't very clearly defined, is a bad/mixed metaphor, or is just plain hard to understand.
My colleagues (developers) laugh when I pause few seconds to think about a proper name for a Java method. But I take it seriously, finding a class or method with proper name is way easier than an ambiguous name> Even a proper name with misspelling is considered bad in my dictionary.
Once in an interview, I asked the guy to write some code, he used variable names like "int ttttt=1;". I never saw him again!!
I think this problem comes up from the somewhat obsolete linear text nature of code. Instead of particular perfect names, we could be using tags. AbstractSingletonFactory is three tags. In some contexts I'd like to know it's a factory, in others I don't.
The bigger problem is we haven't moved on from the simple text representation of code. And this overly simple tool, text, is influencing the way we're thinking here. The Humble Programmer phenomenon in action.
What I like to do is add a variables.txt file to the root of the project as a sort of legend for any unconventional variables I have to name. I try to stay as relevant as possible, but sometimes 'anospic' is better than 'average_numbers_picked'.
In what situation is the name 'anospic' acceptable? In a small scope, 'avg' or 'average' would probably read better. In a larger scope, you'd have to remember that it really means 'average_numbers_picked' each time it appears -- why not just use that name?
Once it's written, I always make the association between 'anospic' and 'average numbers picked' (using that as an example). I try to make the names both human-readable and at least somewhat related (a:average nos:numbers pic:picked).
Keep in mind that I don't do this with client work or anything that might land on the hands of another developer.
There is always another developer - even if it is only some future you. When you pick up that piece of code in three years, hoping to improve or reuse some of it, it will take you three minutes to remember what "anospic" was supposed to mean...
You don't always know from the beginning what's going to land on the hands of another developer, and if something (even a personal project) did shift from you to someone else, you're going to not be very highly incentivized to change things then.
Another possible solution is to put the full variable name in a comment next to the first declaration, in languages with variable declarations. Like this:
double anospic; // average numbers picked
Just raising it as an option; I haven’t actually used that style much myself. With the code I’ve been writing recently I haven’t had any problems so far giving variables names as long as I’d like.
That's a good idea for languages that use that convention (I'd use it with objective-c if xcode didn't make using long variables so easy), but I mostly work with ruby.
I don't buy this argument, but you're in good company holding this belief.
Generally, I find that long names fall into three categories:
1) refactoring opportunities
Use more namespaces, write smaller functions, embrace lexical scoping, etc. You don't need to call something a foo-bar-frobber when it's being used in the foo-bar context, just call it a frobber.
2) line noise
If you're only using the frobber within a two line function, and it is initialized via create-frobber, do you really need to spell it out? f or x will do. If your function grows, rename the variable. Or refactor!
3) an absentee synonym
Maybe you've got a monster-which-attempts-to-defeat-player, when you really need a baddie. Or maybe the word you need doesn't exist yet? It's rare, but sometimes you really do need to invent or repurpose a word.
...
I always feel dirty when I have a long variable name. pg writes about this in On Lisp: long names feel heavier. They might be super cheap to call, but because the name is long, you're less likely to call it.
Furthermore, I much rather adhere to a line length limit, as I find wrapping lines far harder to read than lines with short variable names. Long names makes this harder.
All that said, I almost never use abbreviations other than standin variables like a single letter. The only exception to this is well known abbreviations, like for your most commonly used utility functions or for objects which are extremely common in your problem domain. Such a set of abbreviations becomes a part of the language you share between your compiler and your team.
f or x will do. If your function grows, rename the variable.
Why would you just not give it a reasonable name to begin with? Why not call your variable frobber? The only possible reason for not doing so is that you're using an editor that makes you type it every single time. I insist that calling your Frobber variable frobber is strictly easier to understand than f or x. Why not t? Or sdfsdg? None of them make any sense to someone reading the code, and it increases the amount the reader of the code has to keep in their short-term memory, making it more challenging to understand.
pg writes about this in On Lisp: long names feel heavier. They might be super cheap to call, but because the name is long, you're less likely to call it.
Even if pg wrote it this is patently absurd, and it's endemic in lisp code which contributes greatly to the famously write-only nature of lisp. Again, if you're using a decent editor, there is no reason to call your UserDetails object anything but user-details (or maybe details if you're in a context only relating to users).
> The only possible reason for not doing so is that you're using an editor that makes you type it every single time.
Here's some C:
for (int player_index = 0; player_index < players.length; player_index++) {
int rank = player_index + 1;
render_score(rank, players[player_index].name);
}
Blah. So much for the brain to parse!
for (int p = 0; p < players.length; p++) {
int rank = p + 1;
render_score(rank, players[p].name);
}
I find that much, much easier to parse in my head. The definition of p is available without any backtracking. The word "player" appears five fewer times, reducing confusion with the players array. This is a small case.
If this code were to start to grow, and the definition of p and its usages start to grow appart, it's trivial to replace.... but it's probably better for you to refactor.
That's kind of a pathological case you have there - I don't think many people would argue that loop indices should have long variable names. None of the examples in your post above were loop variables, and what I see a lot of in lisp code is using ps for players, for example. In your example you've used pretty sensible names for players, rank and render_score - why not ps, r and rs respectively?
I'm advocating using full words in general, avoiding long variables (multi-word phrases, adjectives, contextualization, etc), and not being afraid of short variables names if they help, rather than hurt readability.
There are plenty of cases where short names are totally fine. If you've got a module full of string utilities and every single function takes a first argument of type String, why not just call it "s"? Everyone will figure it out in seconds and it will save the mental overhead of differentiating "String" and "string". when that does come up.
>Use more namespaces, write smaller functions, embrace lexical scoping, etc. You don't need to call something a foo-bar-frobber when it's being used in the foo-bar context, just call it a frobber.
I agree with this part and code this way.
>line noise
I generally follow this, it depends on how clear the abbreviation is. If I would need a comment I'll spell out the name instead.
>I always feel dirty when I have a long variable name.
Once it passes a certain size, I do too.
>pg writes about this in On Lisp: long names feel heavier.
This is something I strongly disagree with pg on and I would point to arch as prove that he's wrong here. Arch uses ridiculously short names for everything and I personally can't stand the language. As far as I can tell, the language never got popular despite a decent sized group of people waiting to see what pg would make.
>They might be super cheap to call, but because the name is long, you're less likely to call it.
I think pg arrives at this position because he uses vi. If you have a really nice IDE that does advanced code completion (e.g. I can type "RBV <tab>" for ReallyBigVariable) you'll find that you're no less likely to use a long variable or function name.
>Furthermore, I much rather adhere to a line length limit,
What you say is true but only for variable names. Functions can have longer names, especially those called far away by some client code. For me the rule off thumb is that if you need a comment, you are better off with a longer name.
Good API documentation serves a bit like a glossary too. I'd be worried about the double-maintenance problem, but I suppose it's always there between code and docs. As long as you name things consistently, I can see how it'd be useful.
It's very similar to the phenomenon of notation (including maths, sciences, and, yes, programming languages themselves):
"By relieving the brain of all unnecessary work, a good notation sets it free to concentrate on more advanced problems, and, in effect, increases the mental power of the race." -- P Davis and R Hersh The Mathematical Experience (Boston 1981).
I know that my team spends an inordinate amount of time debatting the names of concepts. We've found that, since we're trying to build something unique, there aren't always words for the concepts we're building. Worse yet, there are words that are close, but not quite right! It's extremely difficult to write any code before everyone can hear a word and reach a common mental image. I've found that if you get the right names, the code is so much easier to write and reason about.
Sometimes though, not finding good names might just be an indication of the newness or non-standardized nature of the situation you are using the name in. Even in natural languages, there are concepts/situations which have names in one language and must be described using sentences in others. Trying to fit something into the currently popular wordlist might then be an artificial requirement.
Choosing the right metaphor in terms of OOP is equally important as well as the understanding the word's meaning and the implication of alternative meanings.
It's an example of using a language's idioms. If you use i in a loop, there's no need to explain yourself since there's a common understanding as to what this means.
If you use i, you're a bad programmer. You can't search for the variable name, you can't make sure it is used correctly everywhere. You're a bad person.
So I've started also using a thesaurus when I code and so far I'm happy with the results. It's been helpful in accurately framing some of the hazier concepts with interesting names. I also revisit the naming much less than I use to, because I'm comforted with the notion that if I couldn't find a better word with a thesaurus, then maybe there isn't an obvious one. I also now use words that are clear, but that I so far neglected to include in my programming vocabulary, because other more prevalent ones would jump in my mind and put a shadow on my naming creativity.
So, if you need some help naming modules, functions, variables and db entities, a thesaurus could help. Though, take care to not abuse of it.