Re: StreamDevice, asynDriver or some other solution?

[Eric Norum <eric@norum.ca>]

Specifying the terminators in the startup script rather then the protocol file is useful when an instrument has both a serial or ‘raw TCP’ interface (which require terminators) and a GPIB or USB-TMC interface (which often do not require terminators since they have another ‘end of message’ mechanism).  Omitting the terminator information from the protocol file allows the same protocol file to be used in both cases.

If the instrument in question has only an interface or interfaces needing terminators, then specifying the terminators in the protocol file is probably the best way to go.

YMMV

On Jul 22, 2014, at 3:45 PM, Rod Nussbaumer <bomr@triumf.ca> wrote:

Bryan:

I completely concur with Mark's assessment that StreamDevice is the optimal approach for an EPICS interface to your device.

I do have a question for Mark, however. In my use of SteamDevice, I typically do not use asynOctetSetInputEos to terminate lines in the asyn driver. Is there some tangible benefit that I am missing by using StreamDevice to specify line terminators? As a developer, it feels more natural to specify the terminator character in the protocol file, where everything else related to the data communication with a device is specified. If there is a good reason to do it int he asyn layer, I will consider doing that way in future systems.

-- 
Eric Norum
eric@norum.ca