Re: QMHRTVRQ weirdness

On 21-Feb-2014 12:38 -0800, James H. H. Lampert wrote:

I just noticed something about QMHRTVRQ, when I have the OPM
debugger active:

It seems that debugger breakpoint messages of the "/* Breakpoint at
Statement/Instruction /25D5 Program VIEWMAIN Recursion level 1 */"
variety are themselves request messages, and as such, I'm getting
them in my new F9 retrieval-stream. But they're evidently filtered
out of the F9 retrieval-streams of system-supplied command lines.

What is the best way to filter these (and perhaps others I don't
know about) out of an F9 retrieval-stream?

AFaIK the QCMD F9=Retrieve does a very simple parse, such that any command string [request] that starts with '/*' [optionally preceded by any number of blanks] can strip beyond just those leading blanks, can strip all data up until and including the following '*/'. Thus if the remaining text after the '*/' equals '', then the request is omitted from the F9=Retrieve results. That algorithm should function without any issue, because of the restriction as indirectly documented in CPD0022 with regard to the asterisk; i.e. essentially, nested comments are not possible.

FWiW: I actually have a NoOp command [actually I named the command NOP] that I use sometimes, specifically to allow the F9=Retrieve to retrieve comments I type in my QCMD or QUSCMDLN command shell. I find value in the comments being logged when composing scripts, but rather than placing the comments on the commands I am scripting, I usually precede or follow those requests with comment lines, and enabling retrieval in consecutive requests is easier than getting to a "Display full" command entry to perform a direct retrieve from the past logged requests. That is to suggest, while a F9=Retrieve performed in QCMD with the cursor in the command entry area, comments will be skipped, but a comment can be retrieve explicitly by positioning the cursor over the specific comment logged in the historical messages.