Any ideas on technique?

[Douglas Kennedy]

Well you can always take the pedestrian approach of wrapping every occurrence. The reason for this thread was to see if there were alternatives. I don't object this pedestrian approach of wrapping in fact I often do this for ordinary projects. Now and again projects have combinatorial complexity in which several devices in real time can and are in complex states. If issues arise I diagnose them with packet sniffers and tracing of the FSM states and stacks ( quasi Zigbee) of the multiple devices. There is a loss of immediacy with the printf but it doesn't affect this application since protocol stacks are dumped after a transceiver interrupt occurs and an error condition is detected.
Anyway the last post takes this thread back to square one so let me return to my original inquiry. Is there a better technique than the last post by asmboy suggests?
With respect I would prefer that that this thread have forward momentum. Anyone wishing to state their prowess ( number of complex projects they have done etc.) or their incredulity that others might do debugging this way is perfectly welcome to start their own thread.

[asmboy]

AGREED!!!! i would like a better simpler way too. And its clear that you have more potential territory to cover with adopting a uniform approach.

my background - for several years at the beginning -- was pure ASM for pic - mostly 28 pins or less - 16F/C parts. A late adopter of CCS was i - using free assembler in MPLAB etc

The primitive debug i showed , grew out of that ,and lazy moi, tagged along into ccs C programs. i've used the core of what i showed with various debug levels by the admittedly crude creation of debug1 debug2 etc - and the pedestrian use of search and replace in my text editor to manage the debug mess this could be viewed as. mostly it is a small load to bear.

that said - if managing a LOT of different - large code space program - does it not come down to strictly following/enforcing ONE of these methods uniformly? Being basically the only ( usually that is) programmer who benefits/suffers from my good/bad programming methods - what i do is simply the path of least resistance, NOT any brilliant insight, for sure.
Of course simpler yet appeals to me , if i could figure it out;-))

[gribas]

Hi Douglas,

see if this works for you:

[sseidman]

I really try to stay in the habit of making drivers that return error codes. That way, when an error code comes up, you can pass it to your error handler. You can make a verbose error handler or a non-verbose one, or a null one, changing it in debug mode.

[asmboy]