You must log in or register to comment.
nope
- <> is hard to read for humans
Not really. <> is unusually pointy among the brackets and comparisons / bitshifts are used in different places than generics are so I’ve never confused them.
- <> is hard to parse for compilers
I guess? Does this meaningfully increase compilation times?
- It makes the uses of brackets confusing and inconsistent
No. A language that uses () for parameter lists, literals and indexing is much more mentally taxing to parse
Also dropping here the list of contrarian views op listed in the next article:
Language Design: Popular, but Wrong
- static members
- properties
- <> for generics
- [] for arrays
- Type ident instead of ident: Type
- having if-then-else and switch/case and a ternary operator
- having both modifiers and annotations
- async/await
- separate namespaces for methods and fields
- method overloading
- namespace declarations doubling as imports
- special syntax for casting
- using cast syntax for things that are not casts
- requiring () for methods without parameters
Yea what is wrong with static members? What do they even mean with “[] for arrays”? Why is that bad? Method overloding is bad? why??
I’m assuming static members are bad because globals are bad
“[] for arrays” is because they want to reserve it for generics once <> is retired
I think the oveloading thing is about the c/cpp thing where you can define the same function multiple times in the same namespace which yeah sucks imo
I mean in c/c++ statics arent really globals, you cant acess the from outsside their scope can you? They just retain their value or am i wrong?
[] for arrays is the thing that has been used forever so why should we not use it annymore?
Overloading is also pretty usefull, overloading class constructors is great. I am not a 40 year experience developer but learning c/c++ i never thought that was so bad.
I have no idea about c/c++ statics, does c even have statics? What kind of a scope could statics even have?
I’m very much novice myself and I never liked the idea of trusting the compiler with figuring out the correct overload and neither do I like not being able to tell which version of a function is being called at a glance. Named constructors ftw
you can constrain functions with c++20 concepts to ensure the compiler is calling the correct function if you’re that worried
I mean the thing with overloading is that your functions should have some difference in the paraameters they take, if you make 3 functions that have the exact same parameters of course you will not be shure what the compiler does(alötho i dont think that it would compile? But i dont think that i have ever done that)
If you have a foo(int x float y) and a foo( int x ) function and you call it with just a x as parameter you can be shure the compiler will call your second function. If the compiler for some reasson tried to use the first foo it would throw a error because it wants a int and a float and you just gave it one int.
I am shure that
Foo(){ static int x =0;
X +=1; Printf(“%d”,); }Foo(); every time foo is called x increments so print will be 1,2,3,4… for every call of foo
Printf(“%d”,x); <- wont work because x cant be acessed here, it is out of scope.
If you read more than just the headings, you’d find out that your objections have been addressed in the article. ;-)
I have and they are not addressed, that’s why I commented as such. How would I know that one of the reasons you think <> are hard to read is because they are used as comparison and bitshift or that you intended () to be indexing syntactic sugar if I hadn’t read them? As for the second, I didn’t think how different languages managed to parse them matters as long as it doesn’t impact compilation times significantly, hence my comment.
deleted by creator
It’s a bit of a change but certainly the right thing to do.
My only disagreement with the article is the get/set stuff. I still want to keep something like the old
container[index]syntax, maybecontainer.[index]to indicate that it’s a form of access. As long as generics go after names, this would not cause ambiguity.What’s a form of access but a function from some index type to some element type?
What’s a struct, but a tuple with some names?
What’s computation but state and a transition function between states?
What’s computation but a set of functions transformed by simple term rewriting?
Let people enjoy their syntax sugar.Wasting a perfectly good pair of brackets on some random function call and then suffering for it in many other places sounds more like syntactic salt.
What? I agree with
function[T]style generics, and would be willing to change the access syntax to something likecontainer.[index], as the dot makes the difference quite clear. Or do you mean the approach to implementing a container or the way the compiler has to transform it into the set operation/mutable access? I didn’t think that was such a problem, and I quite like the way it is done in rust, but that approach may be unavailable to many languages.
Stop using types for variables