我正在从 boost::asio::ip::udp::socket 读取，如下所示：
using boost::asio::ip::udp;
// ...
char recv_buf[128];
udp::endpoint sender_endpoint;
size_t len = socket.receive_from(boost::asio::buffer(recv_buf), sender_endpoint);
现在，这工作得很好，但我现在能够接收的最大字符数是127. 然而，我面临一个问题，因为我需要接受一些长度可能变化很大的数据输入（例如，长度不明确且带有前缀标头）。解决这个问题的方法是动态扩展缓冲区，就像向量一样。是否可以创建一个动态扩展的 boost::asio::buffer 来接受（理论上）无限量的输入并将其存储在容器中？
I'm reading from a boost::asio::ip::udp::socket like this:
using boost::asio::ip::udp;
// ...
char recv_buf[128];
udp::endpoint sender_endpoint;
size_t len = socket.receive_from(boost::asio::buffer(recv_buf), sender_endpoint);
Now, this works perfectly fine, but the maximum amount of characters that I am able to recieve is now 127. However I am facing a problem because I need to accept some data input of which the length can greatly vary (and is not of well-defined length with prefixed headers, for example). A solution to this would be a dynamically expanding buffer, like a vector. Is it possible to create a dynamically expanding boost::asio::buffer to accept (theoretical) infite amounts of input and store it in a container?
发布评论
评论（4）
UDP 数据报大小变化不大：它永远不会大于 65535，离开8 字节标头后可容纳 65,527 字节数据。
UDP datagram size does not vary all that greatly: it will never be greater than 65535, leaving room for 65,527 bytes of data after the 8-byte header.
Boost 1.66.0 添加了
dynamic_buffer，它可以调整对std::string或std::vector的引用：Boost 1.66.0 added
dynamic_bufferwhich can adapt a reference tostd::stringorstd::vector<CharType>:如果您使用较小的缓冲区，您可以通过 *BufferSequences 概念轻松地将它们链接在一起。例如，您可以传入 MutableBufferSequence
接受来自 read(2) 调用的数据或传递 ConstBufferSequence 用于您要写出(2) 的缓冲区列表。也就是说，我倾向于建议在每个方向使用单个缓冲区，因为它可以简化代码（尽管这并不总是可行）。
If you use smaller buffers, you can easily chain them together via the *BufferSequences concepts. For example, you can pass in a MutableBufferSequence
to accept data from a read(2) call or pass a ConstBufferSequence for a list of buffers that you are going to write(2) out. That said, I tend to recommend using a single buffer in each direction because it tends to simplify code (though that's not always possible).
似乎没有任何动态调整大小的规定。不存在也是有道理的。考虑一下会发生什么：
因此，动态大小的可用缓冲区是没有意义的。正如 Cubbi 指出的，无论如何，UDP 数据报的最大大小都较小，因此只需将缓冲区设置为与系统中最大的有效消息一样大即可。
There does not appear to be any provision for dynamic sizing. And it makes sense that there would not be. Think about what would have to happen:
So it does not make sense for there to be a dynamically-sized buffer available. As Cubbi points out, UDP datagrams have a smallish maximum size anyway, so just make your buffer as large as the largest valid message in your system and be done with it.