On 30-Aug-2011 09:21 , Timothy Adair wrote:
Is there a way in CL to set up a monitor that would increment a
counter after each line of code (or each command) is executed?
Just change the non-OS CL statement processor to perform the counting
and any optional action(s) after each line of CL is read [and
processed]. As a CL stream with no programmed logic to skip statements,
that is about the best way to know the total number of statements that
will be processed from start-to-end.
Although presumably CLP was meant instead of using scripted and
interpreted CL.? So more likely... Change from using a compiled CLP to
using a CL statement processor which does that extra work, or change to
use an interpreter like REXX. REXX even has a SOURCELINE() function
which, depending on implementation, could be useful in that regard.
There is also STRDBG with ADDBKP BKPPGM() [Add breakpoing; Breakpoint
program to call] that a CLP could use with static labels or dynamically
generated breakpoints by review of the source. There is little benefit
however, for using static labels over CALLSUBR, coded in the same places
of the source.
Better for a CLP is STRDBG with ADDTRC TRCPGM() [Add trace; Trace
handling program] which for STMT(*ALL) can enable an exit after every
executable CL statement. However the total number of statements would
have to be coded statically or dynamically retrieved or estimated
[optionally cached for each invocation after the first if that were
costly processing], for use in the trace exit-program. The mention of
"batch processing" in the help text of ADDTRC belies the actual effect
from my recollection of my own experience for use interactively.
In either case above using STRDBG, the CLP could be coded to debug
itself, but that is not something most people would like to do. The
debug and trace in the latter case allows the process to be most generic
and independent, esp. if the total number of statements is determined
dynamically.
Regards, Chuck
midrange.com
