Posted 24 September 2023,

This post addresses the problem I have started to see regarding print buffer overflow due to the decreased time (now 50mSec) between environment updates. When running longer field tests, I have started to see many lines where telemetry printout has become garbled, with the next line of printout starting before the previous line has finished, as shown in the accompanying printout

In the above printout, you will notice that the horizontal scrollbar is much smaller than normal, indicating the line lengths are much longer than normal, and the portion that is visible is really really ugly!

So, what to do? I ran a search through my almost 7000 lines of code for ‘gl_pSerPort->printf(‘ to see how big of a problem I’m looking at, and the search came back with some 400+ hits – ouch! One thought would be to put an ‘if’ statement around each printf call and use an ‘EllapsedMillisec’ variable to limit calls to, say, every 100 or 200 msec, but doing that for 400+ calls would get pretty tedious. Another thought is to abstract all these print statements into a function, and have the time spacing internal to the function. So the function call would look something like:

and the function internals would look something like:

The problem with the function internals is the argument list would have to match the contents of ‘formatstr’, which means I would have to have an overload for every unique format, and changing anything would be a hassle, as I would have to find and edit the correct overload – yuck!

OK, more thought needed on this – maybe the ‘if’ statement is the way to go (maybe a macro?)

Another thought that just came up; if I filter all print statements by an elapsed time gate, that might actually filter out important one-time error information – ouch! This potential problem suggests adding a priority value to the ‘LogPrint()’ function, so the function might look like:

And good programming technique would demand that the PRIORITY_CODE have a default value, which (for Arduino programming) would require that every overload have a function declaration at the top of the program – yuck! yuck!

Maybe it would be easier to just put the

around each and every ‘low priority’ print statement (all 400+ of them)? yuck! yuck! yuck!

After thinking about this some more, it occurs to me that this problem may be a lot smaller than I think. The primary culprit is the long telemetry string printed out each WALL_TRACK_UPDATE_INTERVAL_MSEC (currently set to 50mSec). It may well be that I only need to guard this one print statement and 99% of the time that will do the trick. As I find other printout statements that are causing problems, I can treat them as well.

25 September 2023 Update:

I did a search for ‘gl_pSerPort->printf(‘ but bypassed everything in setup() and loop(). Starting in TrackLeftWallOffset() in the ‘while (gl_LastAnomalyCode == ANOMALY_NONE )’ block I put an ‘if(mSecSinceLastTelemetryUpdate > PRINT_INTERVAL_MSEC)’ guard around the call to ‘OutputTelemetryLine(TRACKING_LEFT)’, as shown below:

26 September 2023 Update:

After guarding the ‘OutputTelemetry()’ call in both TrackLeftWallOffset() and TrackRightWallOffset(), I ran another field test, and saw that telemetry output was clean, and now coming in 0.1sec intervals, as intended – yay!

Stay tuned,

Frank