Game Development

« first day (1751 days earlier) ← previous day next day → last day (3284 days later) »

11:01 PM

But it's trading one large convenience on my end, for many inconveniences on the user's end :(

@OMGtechy It's not performance critical, so I have to go with the "most correct" solution

template <class A, class B> void LameCopy(A& dst, B& src) {
	char* D = reinterpret_cast<char*>(&dst);
	char* S = reinterpret_cast<char*>(&src);
	int L = (sizeof(src) < sizeof(dst)) ? sizeof(src) : sizeof(dst);
	while (L--) D[L] = S[L];
	return;
}

OMGtechy

why can't you just use memcpy again?

MickLH

I don't know, I probably can, but it takes void* not char*

OMGtechy

yeah that's the point

where you can void* you can basically have T*

it's the "whatever" pointer

front the days of C where we didn't have templates

MickLH

I know that char* and unsigned char* are the magic ways to be "allowed" to break strict aliasing

I'd have to do a research report to find out if void is lol

TheMuffinCoder

OMGtechy

11:13 PM

Q: When to use a void pointer?

I understand the use of void pointer for malloc implementation. void* malloc ( size_t size ); Can anyone suggest other reasons or provide some scenarios where it is useful in practice. Thanks

@MickLH

@TheMuffinCoder it is the sacrifice we make

TheMuffinCoder

@OMGtechy is this the only reason we go to college then?

to find people who do? :P

OMGtechy

ahahah

I don't know what it's like over there, but over here it's mainly to get drunk and have a piece of paper that says "I know stuff!"

I'd imagine the same

TheMuffinCoder

@OMGtechy xD same here basically

OMGtechy

I never went though

TheMuffinCoder

lol but i do think college will teach a lot

OMGtechy

11:15 PM

I assume when you say college you mean what I call university btw

and I should hope so

TheMuffinCoder

sure! :P

hoping to get to go somewhere good for really cheap

gotta take all the ap's

OMGtechy

I just realised it's been so long since I wrote a sorting function I've literally forgotten how to do it

after a week at work my mind just doesn't want to wake up

soooo I'll just use std::sort

TheMuffinCoder

quora.com/…

am surprised Brendan Eich answered this himself

OMGtechy

niceee

MickLH

11:28 PM

@OMGtechy so........ gcc provides __attribute__((__may_alias__)) lol

to basically do -fno-strict-aliasing on only one variable

OMGtechy

do you care about anything other than gcc?

MickLH

Theoretically I don't have to ;)

OMGtechy

umm ok

OMGtechy

11:41 PM

@MickLH wellllll I've just gone and done some UB of my own; the union thing

for now I cba and know it'll work 99% of the time, and there's a TODO to do it properly

        union
        {
            byte bytes[4];
            uint32_t count;
        } toWrite;

        // strictly speaking this is UB, but it'll do for now
        // whilst most use unions to reinterpret bytes, it's UB!
        // with that said, most implementations do what's expected
        // TODO - correct this
        toWrite.count = static_cast<uint32_t>(lookup.size());

        target.write(toWrite.bytes, 4);

Q: Error. Fix it now or else the world will end

<div id="y"> <button onClick="p3()"> 3 Players </button> </div> <div id="x"> </div> <button id="bla"></button><button class="a" id="whi1"></button><button class="a" id="whi2"></button><button class="a" whi="whi3"></button> <script> var x = document.getElementById('x'); ...

javascript html

MickLH

uint32_t count = lookup.size();
target.write((char*)&count, 4);

I think that's not actually UB

OMGtechy

ah fair point

well it's more defined anyway

I'll do that, ty

Nican

I would also use sizeof(toWrite) instead of 4.

OMGtechy

even better

MickLH

int main() {
    if (sizeof(int) != 4) printf("Warning: Lol ur so screwd mate. GL!\n");
    // program here
}

OMGtechy

11:48 PM

uint32_t count = static_cast<uint32_t>(lookup.size());
target.write(reinterpret_cast<char*>(&count), sizeof(count));

@MickLH well I'm using uint32_t for that exact reason

MickLH

That makes the 4 completely fine

I believe the purpose of stdint.h is to have these fixed size integers consistently

OMGtechy

yeah but using sizeof expresses it more clearly and means I don't break things if I change it to a 64 bit int later

@MickLH yes it is

MickLH

lol... except your file (network?) format

OMGtechy

it's only a toy :P

Nican

Up to the point you decide you want to make toWrite a smaller object to save space/bandwidth/whatever, and have to find all instances of 4 in your code.

OMGtechy

11:50 PM

Nican

Actually, nvm, that would not help much in this case.

MickLH

You still have to find all instances of uint32_t in this scenario

OMGtechy

I don't take endianness into account yet

@MickLH not if I type using data_t = uint32_t

or something like that

MickLH

That's that scenario ;)

OMGtechy

well I only have it in one place at the moment so shh :P

if I have more dictionary values than a uint32_t can hold then wth man

MickLH

11:51 PM

Getting the right answer means correcting your wrong answer before anyone sees ;)

OMGtechy

there's a check in there that's basically "lol no" if you need more than that

MickLH

lol I am fairly sure that using memcpy the way we talked about earlier is well defined but I can't prove it

OMGtechy

At work we have this stuff that formally proves that some software / hardware has no bugs, my reaction was like WTF

I didn't think that was possible

but it turns out you can

MickLH

It seems to come down to the implementation of memcpy itself, as long as memcpy is implemented with char* (or unsigned char*) then I believe it's defined.