Send array of float with TcpClient

Getting Started
1

As per title, I am trying to send an array of float using TcpClient. Is this possible? should I convert it into bytes before? if so, how?
i’ve tried both

client.write(packet);
client.write(&packet, sizeof(count));

but neither works.

any help would be very much appreciated :slight_smile:

2

Hi @marcodat

Have you looked into using a union in C? That is way for floats and bytes to be accessed in the memory locations.

3

A simple client.write((const uint8_t*)array, sizeof(array)) should do too.

But when you say “neither works”, the resulting error message is the first help you get.
And posting the exact error message(s) would be good practice too :wink:

1 Like
4

Thanks for replying.
@bko I don’t understand how unions could help? I use the array of float because ultimately I want to send a Quaternion.

@ScruffR from the error line

test2.cpp:125:37: error: no matching function for call to 'TCPClient::write(float (*)[5], unsigned int)'

                                     ^
test2.cpp:125:37: note: candidates are:
In file included from ./inc/application.h:56:0,
                 from test2.cpp:2:
../wiring/inc/spark_wiring_tcpclient.h:47:17: note: virtual size_t TCPClient::write(uint8_t)

I should have understood unit8_t was the way to go.
is memcpy() a good way to convert floats into unit8_t? I

5

You could use memcpy(), but there is no need since you already have a byte array filled with your data in memory, just tell the function the pointer you provide is pointing to a byte array.
That’s exactly what my suggested type cast does.

For example, you have a float myFArray[5] and by doing this (uint8_t*)myFArray you “fool” the compiler into beliefing the pointer points to an uint8_t[]. And since sizeof(myFArray) returns the number of bytes occupied by your float array (5 elements with 4 byte each = 20), you’ll pass all the bytes making up all your floats into the function.

1 Like
6

amazing, thanks a lot. will try it now

7

@ScruffR it did the job, thank you.

Would you know how I have to deal with photon’s endianness?

8

There are multiple ways on the Photons side and on the remote side, but before you bother with that, first check if both sides don’t already agree there :wink:

If you really have to convert, do it on the side you are most comfortable with.

1 Like