@programminginc - страница 3719

I wouldn't call using a switch in place of if - else if - else a micro optimization. In quite a few cases a switch statement leads to more readable code than a long list of if-else branches. If a compiler can optimise it better (in many cases a switch statement can be optimised better) then all the more why I should use it. The point is that readability of code is more important and a switch statement helps there

VD

nowadays I mostly use dlang+grimoire for that

I did love dlang until they included a garbage collector (that it is optional doesn't matter)

Ludovic 'Archivist'

I did love dlang until they included a garbage collector (that it is optional doesn't matter)

Meh, there is worse things than a GC, I would have one anyway with the scripting language and having one here makes sense

Ludovic 'Archivist'

If dlang didn't have a GC, I would just use C++

Light

Depends on the coding guidelines of company

Company?? Bro, get out of matrix, you should open a company if they dont listen to your demands, i am already making millions by listening to my self (ludo coins)

borealis

thx andrew tate

Ілля

Hello everyone, if I store numeric integer variables of type char (-128...127) or unsigned char (0...255), does it make sense? I understand that this will save memory, since the char type takes only 1 byte, but it will also take time to convert from char to int. Does this method have any other pitfalls? Thx for answers.

Maxim

Hello everyone, if I store numeric integer variables of type char (-128...127) or unsigned char (0...255), does it make sense? I understand that this will save memory, since the char type takes only 1 byte, but it will also take time to convert from char to int. Does this method have any other pitfalls? Thx for answers.

type conversion is not always something that survives the compile phase

borealis

assuming x86, if it survives the compilation, you will need to read the value into a register to operate on it. in case of loading an int, it can be done with a mov. loading a char to a register will instead sign extend it with movsx. both of these take the same time to execute. in other words, using a char as an int incurs no more performance penalty than using an int as an int

Ludovic 'Archivist'

Company?? Bro, get out of matrix, you should open a company if they dont listen to your demands, i am already making millions by listening to my self (ludo coins)

You sound like you are gonna recommend me both a financial placement with a 6% monthly interest rate and "zero risk™️" as well as a funding plan with an 400% interest rate that forces me to share 3% of my company with you

The

Any idea in what case can a std::vector<std::shared_ptr<const std:: string>> have an empty shared ptr as a member.. I'm doing vec.erase(vec.begin()) vec.push_back And at one place func(*vec.front()) As per my thinking vec.erase should always destroy the shared ptr completely correct.. even if doesn't, front and begin should move forward properly..

borealis

youre either pushing a nullptr into the vector or std::move'ing from a shared ptr. also removing from the front of a vector is very expensive, you might want to use std::deque instead

The

youre either pushing a nullptr into the vector or std::move'ing from a shared ptr. also removing from the front of a vector is very expensive, you might want to use std::deque instead

Can you explain me a little why std::move is doing this.. I am not moving the shared ptr.. but while creating it using make_shared, I move the pointee object to it

Chat Boss

Ілля Корнійчук sent a huge message, it has been re-uploaded as a file Hello again, I have two cycles, in the first one I initialize a new variable at each iteration, ..

Ілля

Hello again, I have two cycles, in the first one I initialize a new variable at each iteration, and in the second one the variable is initialized outside the cycle, which option is more efficient? int main() { for (int i = 0; i < 5; i++) { int a = i*2; cout << a; } { int a=0; for (int i = 0; i < 5; i++) { a = i*2; cout << a; } }

Ludovic 'Archivist'

Ілля Корнійчук sent a huge message, it has been re-uploaded as a file Hello again, I have two cycles, in the first one I initialize a new variable at each iteration, ..

Since you are using integers, and integers are "plain old data" those are probably almost exactly identical

Ілля

Rather, the question here is in the logic itself, because instead of the int type, I can use another custom type that may take more time in the constructor. I'm just wondering if the compiler can somehow track and optimize this, or if the programmer has to foresee it?

Pavel

Rather, the question here is in the logic itself, because instead of the int type, I can use another custom type that may take more time in the constructor. I'm just wondering if the compiler can somehow track and optimize this, or if the programmer has to foresee it?

Compiler can do that sometimes, but only if types are simple (don't have custom constructors, that are either complex or defined in another translation unit)

borealis

Can you explain me a little why std::move is doing this.. I am not moving the shared ptr.. but while creating it using make_shared, I move the pointee object to it

this is fine, you must be doing something else that causes the problem

Pavel

Hello again, I have two cycles, in the first one I initialize a new variable at each iteration, and in the second one the variable is initialized outside the cycle, which option is more efficient? int main() { for (int i = 0; i < 5; i++) { int a = i*2; cout << a; } { int a=0; for (int i = 0; i < 5; i++) { a = i*2; cout << a; } }

In this case the result of i*2 will probably be stored in a register in both cases, so the variable will be optimized away anyway And if the code around will be very complex, then it can be that the second variant can be actually less efficient (I assume it can happen that while generating the code for the loop the compiler won't be able to make sure that a won't be used outside of the loop and will decide to store it on stack).

Pavel

But that just thoughts, I'm not sure if that can happen

borealis

Rather, the question here is in the logic itself, because instead of the int type, I can use another custom type that may take more time in the constructor. I'm just wondering if the compiler can somehow track and optimize this, or if the programmer has to foresee it?

the initialization outside of the loop is redundant. if you're going to keep overwriting your previous object in every iteration, you can and probably should initialize it within the loop. if you have an object that is expensive to construct and destruct, it is better to initialize it once, outside of the loop, and keep reusing the same object

borealis

actually, if construction is expensive but the assignment operator is cheap for your custom object, it would be better to construct once outside of the loop and keep assigning to it in every iteration. overall this is a context specific question and there's not a definitive answer without seeing the custom object

FriedRice

There is no difference, the compiler will just put a in a register.😂

FriedRice

But for coding style you probably want to declare a as closed as possible to where you will use a.

\Device\NUL

Company?? Bro, get out of matrix, you should open a company if they dont listen to your demands, i am already making millions by listening to my self (ludo coins)

Lmao. Talk to me again after you already mature enough

不要惹胖虎

the world isn't easy that you want

Adaita (MedTech | Healthcare)

Company?? Bro, get out of matrix, you should open a company if they dont listen to your demands, i am already making millions by listening to my self (ludo coins)

Funny guy. Sounds like you are capable of replacing all the companies in the world. Coding guidelines are followed strictly in many companies for many advantages.

\Device\NUL

Hello everyone, if I store numeric integer variables of type char (-128...127) or unsigned char (0...255), does it make sense? I understand that this will save memory, since the char type takes only 1 byte, but it will also take time to convert from char to int. Does this method have any other pitfalls? Thx for answers.

implicit integer promotion

Light

I wanna send a codebut its little large how can i do that?

Light

(Sending code for evaluation)

Danya🔥

I wanna send a codebut its little large how can i do that?

Use GitHub or GitHub gist

кар карыч

I wanna send a codebut its little large how can i do that?