Hello folks. Does someone know why write to a socket buffer at the full speed of a while loop generate a lot of data loss on client side, and when i insert an sleep(1) right after the server write function, client can read everything without data loss? (Both buffer have the same size)

Hello folks. Does someone know why write to a socket buffer at the full speed of a while loop generate a lot of data loss on client side, and when i insert an sleep(1) right after the server write function, client can read everything without data loss? (Both buffer have the same size)

See what I am doing here to handle such situation. https://github.com/TeaInside/simple_file_transfer/blob/master/users/ammarfaizi2/client.c#L221-L288

The problem is that, write to socket buffer is not always 100% successful. If your buffer is full, your write operation will get error EAGAIN, or it will block if it is blocking socket. If your write length is bigger than available buffer, the data will be written partially, you need to resend the bytes that have not been sent.

The problem is that, write to socket buffer is not always 100% successful. If your buffer is full, your write operation will get error EAGAIN, or it will block if it is blocking socket. If your write length is bigger than available buffer, the data will be written partially, you need to resend the bytes that have not been sent.

Can you explain the step where data is lost ? I dind completely understand it!

Is server buffer consumed for the client part ? So it happens because different speeds. Is that right ?

Ok, i think i got it. So the solution is: if ret <1024 Then: write(fd, buffer+red, buffer_size - red)

Ive inserted a simple header in the first 256bytes, so it need to garantee that it will arrive in the right positions of the client's buffer.

SOCK_STREAM

SOCK_STREAM

Alright, you can guarantee with that header.

Can i send you a message if i have more doubts related to it ?

Send here in the group, you can mention me if you want. I just don't want to discuss it privately.

Hi guys If u do have mql4 course Pls share

I want to write some benchmarks for my project, for example I have complex containers and I want to measure how much time it takes to add/remove stuff in these containers on average in specific situations, to then see a trend if these times will become worse at some point of development. Which frameworks would you suggest, to write such benchmarks?

Google Benchmark maybe?

Is c++ is good for arduino

it is built using C++

string strmulti(string s1, string s2){ string str=""; string mask=s2; while(s2!=""){ int i=1; int a=stoi(s1); int b=stoi(s2.substr((s2.size()-1), 1)); if(s2==mask){ str=to_string(a*b); s2.erase(s2.size()-1); i++; } else{ string temp=to_string(a*b*pow(10,i)); string mask2=str; int tmp=0; str=""; while(temp!="" || mask2!=""){ str=to_string((stoi(temp.substr(temp.size()-1))+stoi(mask2.substr(mask2.size()-1))+tmp)%10)+str; tmp=(stoi(temp.substr(temp.size()-1))+stoi(mask2.substr(mask2.size()-1))+tmp)/10; temp.erase(temp.size()-1); mask2.erase(mask2.size()-1); } i++; } s2.erase(s2.size()-1); } return str } Is this is correct for string multiplication????

There's no such thing as string multiplication

like i made this for factorial function, the problem is, 99! for that i needed string multiplication

string strmulti(string s1, string s2){ string str=""; string mask=s2; while(s2!=""){ int i=1; int a=stoi(s1); int b=stoi(s2.substr((s2.size()-1), 1)); if(s2==mask){ str=to_string(a*b); s2.erase(s2.size()-1); i++; } else{ string temp=to_string(a*b*pow(10,i)); string mask2=str; int tmp=0; str=""; while(temp!="" || mask2!=""){ str=to_string((stoi(temp.substr(temp.size()-1))+stoi(mask2.substr(mask2.size()-1))+tmp)%10)+str; tmp=(stoi(temp.substr(temp.size()-1))+stoi(mask2.substr(mask2.size()-1))+tmp)/10; temp.erase(temp.size()-1); mask2.erase(mask2.size()-1); } i++; } s2.erase(s2.size()-1); } return str } Is this is correct for string multiplication????

Why not test it? Given that it fails for strmulti("1", "1"); I'd say it's wrong

Is there a way to use a different index type for std::variant? By default it's int AFAIK which is 4 bytes and too much for my needs. I would like to use a byte std::uint8_t as index and pack the structure tightly to save more memory...

Yes, there is sure

Except for writing my own variant?! How?

I'm looking into MSVC implementation and it seems like they are choosing the type of the index depending on the amount of types of the variant, but it's not exposed to the user https://pastebin.com/PxM3zDb4

Hmm interesting but my other variant entries are all 8 bytes - the current size is 16 bytes I hoped to reduce it to 12 or 10 by using #pragma pack und an some different index type :/

Can it be related to alignment/packing?

Yeah it is - the whole structure is padded to 16 bytes which is okay in most cases but in my case it takes 29 GB of RAM to store it because it's over 1 billion entries 😂 I've already compressed some pointers using 32-bit indices but it would still be a big win to save 4 or 6 bytes...

did you try something like #pragma pack(push, 1) and #pragma pack(push, 1) on MSVC? I guess it may help to pack it tighter

Yeah I already have (works with Clang, GCC too) but there was no difference in size :/

:(

So it depends on what you need. If you are writing a lot of data and performance is critical don't use endl

#include<iostream> using namespace std; int main() { int A[2][3]={{2,5,9},{7,3,6}}; int B[2][3]={{6,3,4},{9,5,2}}; int C[2][3]; for(int i=0;i<2;i++) { for(int j=0;j<3;j++) { C[i][j]=A[i][j] +B[i][j]; } } for(int i=0;i<2;i++) { for(int j=0;j<3;j++) { cout<<C[i][j]; }cout<<endl; } }

Pls anyone

You tell here what's the error you got when you compile/ ran.