I have no experience with single-board processor machines, but it seems to me that simple is the way to go on that side; simple probably means a small executable developed in LabWindows/DOS, Borland C, BASIC, or whatever. (As opposed to elegant, which would be exposing the DOS filesystem to the network through some proprietary protocol like Banyan VINES, etc.) You can program the LabVIEW side of things for whatever analysis flexibility is needed.
That simple executable on the DOS side would:
1. Listen at a particular port for a TCP connection from the LabVIEW client
2. Once the connection is made, spew the contents of the file over on a regular basis; perhaps use some "diff" logic to spew over only the changes
3. If the connection is dropped, create a new
listener
On the LabVIEW side, you'd have a simple app along the lines of the Data Client.vi/Data Server.vi Networking examples.
This isn't much specific help, but I bet there are people out there familiar enough with networking/socket programming in common tools like C and BASIC to rip off the DOS executable without much trouble.
Regards,
John Lum
National Instruments