JewKiller 3000 posted:

for primitives it makes a copy of the value on the callee's stack. for objects it does NOT copy the object, it makes a copy of the pointer to the original heap object on the callee's stack. if the object has mutable state and the callee invokes methods that modify it, the caller WILL see those changes reflected on his side, since the new pointer copy still points to the same heap object.
however, if you take a different object and assign it to your copy of the pointer in the callee, the caller WILL NOT see that change, because the pointer was passed by value (create a copy) and not by reference (new pointer aliases original pointer). if you want a deep copy of the object for the callee that won't affect the caller, you want to use a copy constructor.

i know i need to stop posting about this but people won't stop being wrong so i can't.

Zlodo posted:

see ok i admit i tend to poo poo on java without knowing that much about it (except i dont like the language and its slow), yospos bitch and all that but hold on:

a function can mutate an object passed to it? but you can overwrite it locally with another object? thats really not what id call "pass by value", rather its exactly what passing by pointer in c++ accomplishes: you have a reference to an object, but you can make it refer to a different object without the caller noticing it.
(whereas passing by value in c++ makes anything done to the object completely invisible to the caller)

Zlodo posted:

its much closer to "pass by reference" than to "pass by value" tbh. earlier you said passing by ref is a bad idea because you cant locally reason about whether a function that you call will modify your object but java has an even worse version of that particular problem: you need to actually read the functions code to see if it might invoke methods that mutate your object

Zlodo posted:

whereas in c++ if the signature of a function says "Butt a_butt" (pass by value) or "const Butt& a_butt" (const reference) you know that the function is not going to modify it. if it says "Butt& a_butt" (non-const reference) you know its likely to modify it (because otherwise it'd be a const ref).

JewKiller 3000 posted:

i know i need to stop posting about this but people won't stop being wrong so i can't.

JewKiller 3000 posted:

java is doing exactly what passing a pointer in c++ is doing. the callee can make it refer to a different object without the caller noticing. that's why it isn't pass by reference. if i have the local variables Object x = a and Object y = b, i cannot call a function swap(x, y) which results in my local state changing to x = b and y = a. pass by reference would let me write that swap function (and you can write it in C++)


that's true! mutability sucks a fat one in general


yeah java does not have the const-correctness machinery of C++. there's final but it doesn't work exactly the same way

FamDav posted:

i think the problem is youre really hung up on java being pass-by-value as an explicitly awesome thing when people are like "but brah i can still mutate what that object reference points to" and youre all like "but you can't mutate the reference!" and everyone is like who loving cares.

FamDav posted:

i think the problem is youre really hung up on java being pass-by-value as an explicitly awesome thing when people are like "but brah i can still mutate what that object reference points to" and youre all like "but you can't mutate the reference!" and everyone is like who loving cares.

PleasingFungus posted:

but it's still not pass-by-reference.

I mean, you can say it's poo poo, and I'll agree. but it's not pass-by-reference.

saying 'java is poo poo because it uses pass-by-reference semantics!' is a good way to have an argument that is completely different from the one you (presumably) intended to.

this is why words are important sometimes.

JewKiller 3000 posted:

it's another thing you have to worry about.

Mido posted:

python is duck typed



ducks are cool


ergo


python is a messy pile of poop, never raise ducks

Deacon of Delicious posted:

so what does everyone think about semantic whitespace/brace styles/strong vs weak typing/global variables/spaces vs tabs/does java pass by reference or value/ide vs text editor

OBAMA BIN LinkedIn posted:

can we just agree that Java not having const is a travesty and also that Java is a poo poo language and that c# owns

OBAMA BIN LinkedIn posted:

can we just agree that Java not having const is a travesty and also that Java is a poo poo language and that python owns

OBAMA BIN LinkedIn posted:

can we just agree that Java not having const is a travesty and also that all programming languages are poo poo and that leaving work and going to do other poo poo owns

gucci void main posted:

OBAMA BIN LinkedIn posted:

can we just agree that Java not having const is a travesty and also that Java is a poo poo language and that c++ owns

Hard NOP Life posted:

I actually don't understand how const works in C++ since I'm a javatard. How does adding the keyword const prevent the caller from mutating the objects state or is is just a convention?

Nomnom Cookie posted:

Const references aren't worthwhile IMO. Good code won't mutate params and bad code will, and const doesn't change that. The right thing is to make objects logically immutable as much as possible, which const doesn't help much with. poo poo, C++11 had to introduce mutable because of the difference between a const object and an immutable one.

FamDav posted:

mutable has always existed

Nomnom Cookie posted:

You mean callee mutating the object's state? It doesn't! If the callee wants to mutate the object they just won't declare the param as const.

Nomnom Cookie posted:

I stand corrected. Const is still a bad idea, like checked exceptions.

FamDav posted:

the two interesting arguments about const/mutable keywords:

1) should you ever be allowed to mutate something
2) should you be allowed to mutate by default

Malcolm XML posted:

1) No
2) See 1)

FamDav posted:

the two interesting arguments about const/mutable keywords:

1) should you ever be allowed to mutate something
2) should you be allowed to mutate by default

Shaggar posted:

checked exceptions are good

Hard NOP Life posted:

I actually don't understand how const works in C++ since I'm a javatard. How does adding the keyword const prevent the caller from mutating the objects state or is is just a convention?

Nomnom Cookie posted:

Nope they're bad. Like const, just whips & chains with no reacharound.

FamDav posted:

the two interesting arguments about const/mutable keywords:

1) should you ever be allowed to mutate something
2) should you be allowed to mutate by default

Condiv posted:

const int* const Method3(const int* const&) const;

what's hard to understand about that?

it would be neat if final in java (and val in scala) trickled down to the members of objects so instead of having a final reference to an object, the object becomes immutable.

Nomnom Cookie posted:

You're conflating reference mutation with value mutation. If the object is supposed to be immutable, then it will be written that way. If it's supposed to be mutable, then why the hell are you trying to make it immutable? Const StringBuilder is a ridiculous notion.