library throws SIGPIPE when remote system closes connection

[kyle-github]

This may be related to #64.

When the remote PLC closes the TCP connection, the library code still attempts to call write() and that causes a SIGPIPE on Linux.

This is difficult to trigger because of buffering.

The work around, for Linux, is to use send() instead of write() with MSG_NOSIGNAL as the flags argument. Unclear if there is a similar feature for Windows.

Does this work on BSD-like systems like FreeBSD or macOS?

[kyle-github]

This may be related to user GBT's problem reported on the forum.

[kyle-github]

This should be closed. Testing seems to prove that this is now fixed. However, it is difficult to create SIGPIPE errors on demand.