How to Robustly Determine if JSL Code Runs Successfully or Generates an Error?

russ_wolfinger

JSL Gurus,

Working on an update for Assistant and need a way to definitively determine if an arbitrary snippet of LLM-generated JSL runs successfully.   I am currently using the following:

ret = trim(logCapture(eval(parse(jslcode))));

This grabs the log just fine, but a lot of times there is an error and the log text does not actually contain "ERROR" and I am currently stuck chasing down common error message keywords or phrases like "does not contain" or "unexpected".

Is there any clean way or hidden flag available for telling if the JSL ran successfully?   Or maybe you have a better recommendation in general?

Thanks much and Happy Holidays!

Russ

Ben_BarrIngh

Hi @russ_wolfinger 

Could you use the debugger to highlight any failures in the JSL? I know from my rugged JSL experience that sometimes some errors may not even write to the log past "does not contain", "unable", "unexpected" - when debugging I've typically just wrote for sanity checkers to ensure the resulting window or result even exists, ie trying a get text or get script to see if the platform has executed as expected?

 

gb=Graph Builder(
	Variables( X( :height ), Y( :weight ) ),
	Elements( Points( X, Y, Legend( 3 ) ), Smoother( X, Y, Legend( 4 ) ) )
);

gbcontent=Report(gb)<<get text;

//If the graph falls to default form without an X and Y, then it will return "Title", if not then success
if(gbcontent=="Title", "error","success");