BinaryByter
😂
Ludovic 'Archivist'
Sure but it makes the whole constexpr point wrong
No, nothing prevents you from implementing your buffered ostream that can support constexpr writes
Anonymous
excuse me it seems like you just drove into a impasse
Ludovic 'Archivist'
doesnt a buffer imply memory allocation?
Have you heard of our lord and savior the stack?
BinaryByter
BinaryByter
also it bloats the executable size lol
BinaryByter
also it can cause stackoverflow on gcc lol
Ludovic 'Archivist'
thats still not compile time
No, but it is optimizable at compile time
Ludovic 'Archivist'
also it can cause stackoverflow on gcc lol
Only if you write like a half-sliced pigeon
BinaryByter
Only if you write like a half-sliced pigeon
yea, I agree. Gcc is a half sliced pigeon
Ludovic 'Archivist'
how though?
Generators will be optimized into memory copies
BinaryByter
compile time memory copies? 🤨
Ludovic 'Archivist'
Which is exactly what you want
Ludovic 'Archivist'
compile time memory copies? 🤨
You cannot do compile time IO
Dima
Lol compile time io
Dima
Lol this is strange
BinaryByter
You cannot do compile time IO
tahts why I added the raised eyebrows 😉
Ludovic 'Archivist'
tahts why I added the raised eyebrows 😉
Even if you use templates, you will have things optimized to memory copies
BinaryByter
because you are telling me that the compile time thing will be optimized to a non compile time thing
Anonymous
apparently he thinks at the of an impasse he's greeted with a bunch of baguettes
Anonymous
hint: its a wall!
Mihail
because you are telling me that the compile time thing will be optimized to a non compile time thing
it'll be optimized to a fast runtime thing is what he's saying
Mihail
yes
Ludovic 'Archivist'
apparently he thinks at the of an impasse he's greeted with a bunch of baguettes
Could you stop qualifying me by a cliché as it is the most moronic behavior I can imagine?
BinaryByter
+ a string concatenation is just a memory copy too lol
Anonymous
but I'm a moron?
Ludovic 'Archivist'
thats not compile time though
That is NEVER compile time, that is AS GOOD AS IT GETS
Anonymous
you're the smart one so why'd I act like a non-moron?
Ludovic 'Archivist'
but I'm a moron?
Not for me to judge
Anonymous
who'd I be to let you fall on your nose
BinaryByter
BinaryByter
I just said it for the nice silence 😂
BinaryByter
he isn't
BinaryByter
though I agree, lets cool down and stop insulting eachother
Ludovic 'Archivist'
thats not compile time though
Also if you think a template can completely optimize IO at compile time, I have sadly wasted 2 years teaching you
Anonymous
you did, because wim teached him better in 2 months
Anonymous
you diss and agree?
Ludovic 'Archivist'
you did, because wim teached him better in 2 months
Wim teached him such a lost of moronic stuff also lol
BinaryByter
Wim has brought me to look further than just the C++ circlejerk
Anonymous
isn't this the c++ circlejerk the asm circlejerk gets their memes out of?
Wim
I didn't do anything?
Anonymous
you did!
Wim
I just pointed out to look at compiler output/debuggers?
BinaryByter
😂
Ludovic 'Archivist'
I dissagree!
Mmmh I have yet to see the difference between the handrolled asm memcpy and the std::copy in terms of performance kiddo
Wim
damn, why am I in this discussion? 😂
Wim
Mmmh I have yet to see the difference between the handrolled asm memcpy and the std::copy in terms of performance kiddo
So writing it by hand, you're a kiddo, writing it in C and having no control over it actually makes you a pro or did I just interpret that wrong?
Wim
While your C compiler will convert it to ML anyway when compiling..
BinaryByter
having string operations to do it already justifies doing it in asm 😂
Ludovic 'Archivist'
So writing it by hand, you're a kiddo, writing it in C and having no control over it actually makes you a pro or did I just interpret that wrong?
No, writting the code for every asm available to you while a compiler generates the very exact same code from C or C++ makes you a moron
Ludovic 'Archivist'
i doubt that the compiler generates the very exact same code
True, it may generate more suited code on a CPU you did not plan for or use a better reordering than you too
Mihail
i doubt that the compiler generates the very exact same code
it doesn't it actaually looks at context such as platform first
Wim
Its highly dependant on your compilers codegenerator and optimisation flags as well
Ludovic 'Archivist'
And if it doesn't, maybe you should very exactly change the compiler
Mihail
or it might generate worse code :D
does that might outweight all the other pros?
Anonymous
^
Anonymous
^
Anonymous
^
Ludovic 'Archivist'
Mihail
^_^
BinaryByter
does that might outweight all the other pros?
you can quite easily write a solid implementation of memcpy that is compatible with every 8680 compatible CPU
BinaryByter
Now add a few optimizations for some platforms
Wim
i8680?! I never saw those?
BinaryByter
i8680?! I never saw those?
my grandma invented them!
Mihail
you can quite easily write a solid implementation of memcpy that is compatible with every 8680 compatible CPU
but the compiler might have something better and the compiler isn't restricted to i386
Ludovic 'Archivist'
But like I said, if you found an optimization that is not implemented in your compiler... Don't go asm, download the compiler's source and add the frigging implementation
Wim
If automatically generated code is better than hand-written code, you suck at handwriting code