Sending data over the ethernet using server.write()

@gameworn Break the number into two chunks. If you can only send bytes, then get the number into a 16 bit variable, then use bit masks and shift to isolate the low order 8 bits and the high order 8 bits. Send two bytes then reverse the procedure on the other side.

@gameworn

If you can send chars or string, why not just send over the text form of the number. You can then decode it back to float after the transfer if you need to.

@gameworn - I don't use most of the hardware or libraries you are using, but... I'm betting they are using the Stream libraries of the base Aduion code.  So, I bet server.write() is really Serial.write().

https://www.arduino.cc/reference/en/language/functions/communication/serial/write/

It says you can write a buffer.  This means you can send the binary form of any variable IF YOU PREFER and know how to handle it on the other end.  For instance if you wanted to send a float or double... you could use:

float pi = 3.14159265;
server.write((byte*) &pi, sizeof(float));

This will send out the binary version of the floating point structure.  Some things that may look odd:

• The & before the variable says to pass the address of memory location.
• The (byte*) is called type casting and its so the compiler doesn't complain.
• sizeof() function knows how big a float is and will send 4 bytes.

This is very useful because it can be used for whole, arbitrary structures also.

struct Person
{
  char first[16];
  char last[16];
  u8 age;
  float weight;
  double netWorth;
};

Person who = { "Killroy", "Was Here", 42, 0.45666, 23.3466E19 };

void setup() 
{
     Server.write((byte*) &who, sizeof(Person));    
}

void loop()
{
}

@inq Am I misunderstanding? How is his laptop going to access the memory of his arduino?

Posted by: @inq

It says you can write a buffer.  This means you can send the binary form of any variable IF YOU PREFER and know how to handle it on the other end.  For instance if you wanted to send a float or double... you could use:

float pi = 3.14159265;
server.write((byte*) &pi, sizeof(float));

It would be better to write this as ...

server.write( (byte*) &pi, sizeof(pi) );

in case the datatype of variable pi is ever changed.

@will Unless I have completely lost my marbles, &pi means the address of pi. You think the laptop can read a memory location in an arduino? I know I have been away from code for a while, but I think you mean pi.

@zander 

Huh ? I assumed that this was the sample from Inq's post above where pi is declared, assigned a value and then transferred via the server.write() command. I'm assuming that the server.write command is issued from the same machine as the declaration and assignment are made.

The command is just saying "send sizeof(pi) (i.e. this many) bytes starting from this address (&pi, i.e. the location of the value pi) to an available receiver (if one exists and is listening)". I see the command as always working with local values only.

@will I do see two ways of looking at this. I read it as I am sending you sizeof(pi) bytes that is the address(&) of the variable pi. I don't see a prototype for the call so I assumed normal pass by value syntax.

@zander 

OK, but this command will pass down the four bytes (in this case) of the 32bit value for pi, NOT it's address. Is that where we differ ?

@will It depends on the implementation of the server.write. We are both assuming facts not in evidence. I am not talking about @Inq's code, that's just a made up example, it's the OP's code we did NOT see. What he said is he can't send a float, this suggests the write statement expects a single byte. The OP said

I believe the byte type limits this.

Since he got correct data sending a 148 value and incorrect sending 1485 that strongly suggests the value is passed as value and not reference.

NOTE: I recognize you were only responding to Inq.

@will I went and found (I think) the api doc, if he specifies a size as the second argument, then indeed the buffer can be almost any size and the first argument is indeed an address, most often of the form buffer where buffer is char[size] and size is then the value passed as the 2nd argument Otherwise as he seems to be doing perhaps because he doesn't know about the 2nd form is just passing a single char/byte. YES that makes Inq's example correct, but I wanted to make sure the OP got the full message.

@zander 

OK, to limit the confusion, I went back to the documentation at 

https://processing.org/reference/libraries/net/Server_write_.html

Setting aside Inq's supposition that this is equivalent to Serial.write(), it appears that the command will accept an int, a byte array or a String. Given that, I'll stand by my original suggestion that the value be sent as a string (with embedded decimal point) and converted back to a numeric type if/as required.

@will I think 3 of the 4 of us understand. Now if @gameworn would rejoin the conversation his problem would be solved. In this case a formatted string is good enough, performance is not an issue.

Posted by: @zander

@inq Am I misunderstanding? How is his laptop going to access the memory of his arduino?

I believe so... 

Posted by: @gameworn

 How do I send a large number, or maybe a float number using server.write() from the Mega to Processing.

The laptop does not access the Arduino.  These server based libraries typically set up a TCP connection and inherit the Stream class to actually handle the lower level sending as if it is saving to a file (which also inherits the Stream class).   The code on the Arduino writes data to the stream and the laptop side pulls it off the stream and uses it.

Posted by: @gameworn

laptop via an ethernet cable and into a software package called "Processing".

I'm assuming that this "Processing" program can be configured to understand that the next four bytes coming in represents the float number @gameworm wants.   What it does from that point... I have no clue.

VBR,

Inq