Re: Variable ontology (was: Q: Pointers in TCl)
     _________________________________________________________________
   
From           tom_holroyd@yahoo.com
Organization   Deja News - The Leader in Internet Discussion
Date           Thu, 02 Apr 1998 23:08:03 -0600
Newsgroups     comp.lang.tcl
Message-ID     <6g1qrj$jna$1@nnrp1.dejanews.com>
References     1 2 3 4 5
     _________________________________________________________________
   
In article <35237283.1151@cnet.francetelecom.fr>,
  Alexandre Ferrieux <alexandre.ferrieux@cnet.francetelecom.fr> wrote:
>
> Cameron Laird wrote:
> > Who uses subst?
> > confession:  I'm eval-intoxicated
>
> Confession: I'm, too.
>
> Reason: either you disable one of the three substitutions (\,$,[) and
> you violate the symmetrical spirit of the shell, or you do them all at
> once, and you are blatantly redundant with 'eval'...

Right.  Here's an example from a front-end to a graph program
that takes statements like
        label x "data for the first <last_date - first_date> months"
where the dates are variables in the app.  The label proc does a
        calc args
with:

proc calc {argName} {
    upvar $argName args
    regsub -all {<([^>]*)>} $args {[calc1 "\1"]} x
    set args [subst -noback -novar $x]
}

proc calc1 {x} {
    regsub -all {([a-zA-Z][a-zA-Z0-9]*)} $x {[getFromApp \1]} y
    return [expr $y]
}

So, it just uses command substitution after turning <stuff> into
[calc1 stuff], which is a bit like doing a macro..

-----== Posted via Deja News, The Leader in Internet Discussion ==-----
http://www.dejanews.com/   Now offering spam-free web-based newsreading