Fixes#115
The serial out node does a JSON.stringify if the payload is an object. This was incorrectly being applied to Buffer objects, causing the output seen in issue #115.
The Buffer is now passed through as-is (with the newline appended if so configured).
Partial fix to Issue #111
This doesn't fix the screaming loop issue if you try to write to an unplugged
serial port - but does attempt to fix the not retrying to reconnect part of
the issue.
Both were introduced by changes to the underlying serialport npm.
Addresses Issue #104
adds the option to re-use the character used the split input into lineson input as an append to every line sent out to the serial port.
The previous fix to not clone the req/res objects introduced a bug where the req/res objects were lost from all but the first clone made out of a particular node.
Part of #85
A function that returns the same message to multiple outputs, where each output is wired to at most one node was not having its messages cloned due to the change I made in #85 to be slightly more efficient. By returning the same message to each output, cross pollution was possible.
Closes Issue #96
Imap node now logs (in console) if it finds duplicate email on read or not.
It doesn't send on duplicates. It also now DOES fire once on startup AND re-deploy to populate anything downstream... at least that is the excuse for the change.
Signed-off-by: Dave C-J <dave@conway-jones.co.uk>
This allows a websocket-in node to receive data, process it in a flow
and then send it back to the originating websocket client via a
websocket-out node.
Alternative implementation, closes#42
The username/password as not stored in the main flow file for security reasons;
they are stored in the adjacent credentials file. This does mean an extra step
to importing an MQTT node, as the user has to manually edit it to re-add username
and password if needed.
Fixes#85
As well as adding deep-clone (via the new dependency on the 'clone' module), we no longer clone the message if there is a single recipient. This makes simple node-to-node flows more efficient.
I've done some simple profiling using process.hrtime to time how long the Node.send function takes, and at best, this change is neutral to performance.
