Line numbers and debugging verbosity

Line numbers and debugging verbosity

[Bart Schaefer]

On Fri, Apr 6, 2018 at 9:29 AM, Ray Andrews <rayandrews@eastlink.ca> wrote:
>
> What I'm actually trying to do is reduce verbosity of a function by
> redefining various message printer functions as null.  It works fine with
> functions, but if the message printers are aliases (which seem to be the
> only way to get:  ${(%):-%x %I} ... line information printed, it seems that
> can't be done in a function) ... then it goes sour.

On Sun, Apr 8, 2018 at 3:38 PM, Ray Andrews <rayandrews@eastlink.ca> wrote:
>
> Is there a new strategy that might work?  Nullify the function and/or the
> alias?  Really, just some way of stopping messages from printing at all?

There are several possible approaches.  For one, you could stop using
your own "echo"/"print" statements and instead turn tracing on and off
for the surrounding function:

  functions -T test1

If you want to keep printing your own messages instead of using the
xtrace facility, let's get you away from aliases.  The information for
%x and %l are in the $funcfiletrace variable, and will give you the
result you want from inside a function.

  warningmsg() {
    print -n "${funcfiletrace[1]}: " && magline "$*"
  }

For another approach, you had a perfectly good example:

>     [[ "$vverbose" < 3 ]] &&
>     {
>         warningmsg () { ; }
>      }

Just put a similar test in the definition of "warningmsg" e.g.

  warningmsg() {
    (( $dbg )) &&
    print -nr - "${funcfiletrace[1]}: " && magline "$*"
  }

Now you don't have to modify the function, you just assign dbg=1 or dbg=0.

Re: Line numbers and debugging verbosity

[Ray Andrews]

On 08/04/18 10:19 PM, Bart Schaefer wrote:
> functions -T test1 

Not working yet, but I know what to research, so it will work shorty.  
Anyway, the idea is not a full debug, just some helpful indication where 
a message comes from.  I can't stand it when I see some message and have 
no idea who is sending it to me, thereottabealaw!
>    warningmsg() {
>      print -n "${funcfiletrace[1]}: " && magline "$*"
>    }

Marvellous.  Perfect, right out of the box, no need for any alias at 
all.  Why don't they know about this on the internet?  There's many 
queries about line numbers, but the answer is always the alias.
> Just put a similar test in the definition of "warningmsg" e.g.
>
>    warningmsg() {
>      (( $dbg )) &&
>      print -nr - "${funcfiletrace[1]}: " && magline "$*"
>    }
>
> Now you don't have to modify the function, you just assign dbg=1 or dbg=0.

Sure, *once* we're back with functions, it's easy.   Once any need for 
an alias to capture the line numbers is gone, all problems vanish.  
Ironically, last night I figured out how to make it work with aliases 
too, but the above is vastly better anyway.   It's pleasing to know that 
the extra strictness of '5.5 has not really robbed me of any 
functionality, just stopped me from doing things a way that I know 
perfectly well is bogus.