在 tcpip 连接中 fwrite 跟 fread 不同？ [matlab]

Question

在编写一个类的过程中将两个matlab实例连接在一起。这些实例将位于不同的计算机上，但我目前正在一台计算机上进行测试。

目前，我能够在两个 matlab 之间建立连接，并且能够在它们之间发送/接收消息。

代码：

classdef connectcompstogether<handle
    properties
        serverIP
        clientIP
        tcpipServer
        tcpipClient
        Port = 4000;
        bsize = 8;
        Message
    end

    methods
        function gh = connectcompstogether(~)
            % gh.serverIP = '127.0.0.1';
            gh.serverIP = 'localhost';
            gh.clientIP = '0.0.0.0';
        end

        function SetupServer(gh)
            gh.tcpipServer = tcpip(gh.clientIP,gh.Port,'NetworkRole','Server');
            set(gh.tcpipServer,'OutputBufferSize',gh.bsize);
            fopen(gh.tcpipServer);
            display('Established Connection')
        end

        function SetupClient(gh)
            gh.tcpipClient = tcpip(gh.serverIP,gh.Port,'NetworkRole','Client');
            set(gh.tcpipClient,'InputBufferSize',gh.bsize);
            set(gh.tcpipClient,'Timeout',30);
            fopen(gh.tcpipClient);
            display('Established Connection')
        end
        function CloseClient(gh)
            fclose(gh.tcpipClient);
        end
    end
    methods
        function sendmessage(gh,message)
            fwrite(gh.tcpipServer,message,'double');
        end

        function recmessage(gh)
            gh.Message = fread(gh.tcpipClient,gh.bsize);
        end
    end
end

matlab 1

gh = connectcompstogether;
gh.SetupServer();
gh.sendmessage(555);

matlab 2

gh = connectcompstogether;
gh.SetupClient();
gh.recmessage();

发送的消息是一个 8 位双精度 555。 然而，当查看收到的消息时，它原来是一个矩阵

64
129
88

不明白发生了什么，因为我一直关注的examples 没有这个问题。

并添加上下文。我正在尝试通过 TCP-IP 连接 2 个 matlab，这样我就可以控制一个实例与另一个实例。我的计划是让第二个 matlab 等待命令代码并在第一个 matlab 请求时执行指定的函数。

Answer 1

tcpip/fread默认精度为uchar，所以默认fread会输出一个8位无符号整数的列数组。 p>

您要么需要指定预期的双精度：

%size divided by 8, as one 8-byte value is expected rather than 8 1-byte values
gh.Message = fread(gh.tcpipClient,gh.bsize/8,'double');

或typecast uint8 数组为双精度：

rawMessage = fread(gh.tcpipClient,gh.bsize); %implicit: rawMessage is read in 'uchar' format 
% cast rawMessage as uint8 (so that each value is stored on a single byte in memory, cancel MATLAB automatic cast to double)
% then typecast to double (tell MATLAB to re-interpret the bytes in memory as double-precision floats)
% a byteswap is necessary as bytes are sent in big-endian order while native endianness for most machines is little-endian
gh.Message = swapbytes(typecast(uint8(rawMessage),'double'));