[ESS-bugs] tracebug problem: debug() --> from [2] to [36] !!!

Martin Maechler maechler at stat.math.ethz.ch
Fri Oct 25 09:14:08 CEST 2013


I'm debugging an S3 method for "dummy.coef",
I'm working inside *R* exclusively.

Earlier I had

> debug(robustbase:::dummy.coef.lmrob)

and then called  dummy.coef(<lmrob object>)
and hence entered the debugger correctly.

Now watch this

 Browse[2]> 
 Browse[2]> debug(NextMethod)
 Browse[2]> rr <- NextMethod("dummy.coef", object, ...)
 debugging in: NextMethod("dummy.coef", object, ...)
 debug: .Internal(NextMethod(generic, object, ...))
 Browse[36]> 

Note the '[36]' !!!!
I'm 36 levels (yes, and for some reaons the levels are always
       	      "multiplied by 2", so it is ``only'' 18 levels ...)
inside a call even though I only entered one
function.

And indeed, it's no joke:
I can painfully go back level after level :

Browse[36]> c
Error in NextMethod("dummy.coef", object, ...) : object not specified
Browse[34]> c
exiting from: NextMethod("[")
debugging in: NextMethod(.Generic)
debug: .Internal(NextMethod(generic, object, ...))
Browse[34]> c
exiting from: NextMethod(.Generic)
(list "base" '(("generic" . "NULL") ("object" . "NULL") ("..." . "")) '("generic" "object" "..."))
Browse[32]> c
exiting from: NextMethod("[")
debugging in: NextMethod(.Generic)
debug: .Internal(NextMethod(generic, object, ...))
Browse[32]> c
exiting from: NextMethod(.Generic)
(list "base" '(("generic" . "NULL") ("object" . "NULL") ("..." . "")) '("generic" "object" "..."))
Browse[30]> c

This is really ess-tracebug's failure and does not happen if I
use R outside ESS.

I've -- quite rarely -- seen this before, but not recently, so I
had thought this was a temporary glitch in ess-tracebug, but it
seems not.

Vitalie ?



More information about the ESS-bugs mailing list