Re: Request for comments: command line macros

[Juergen W. <wie...@fr...>]

Am Mittwoch, 13. Oktober 2004 20:15 schrieb Ethan Merritt:
> > I was expecting more feedback, but maybe "no news = good news".
> > Yes, the core code can probably go into cvs.  The add-ons
> > should get more discussion. I'll start separate threads for
> > each.
>
> This is a re-implementation of the "userstrings" patch I made
> quite a while back for version 3.8 and still available for 4.0.
> I know people are using this, because they send me Email.
>
> The only issue raised about the earlier patch was that it
> used the symbol "$" for a new purpose, possibly causing
> confusion with multiple existing uses.  So for this new version
> of the patch I have substituted the symbol "@", which is not
> currently accepted by the gnuplot parser at all.
>
> Syntax:
> 	myformat = " 1:2:($4-$5) with lines lt palette"
> 	plot "foo" using @myformat
>
> The parsing code performs a string substitution before
> interpreting the command line, yielding
> 	plot "foo" using  1:2:($4-$5) with lines lt palette
>
> Since @ was not previously legal at all, I can't think how
> this could break any existing scripts or usage.
> But as always, comments are welcome.

It can break scripts because macros are evaluated in "!"- and
backtic shell commands:

  gmx = "foo"
  print `echo us...@gm...`
  ! echo us...@gm...

While the latter doesn't seem to be useful (you can use the system
command instead) the macro expansion in backtics might have some
use.  Consider

  file = "input.dat"
  value = `some-command @file`

I would really appreciate this functionality.  But it is some kind
of inconsistent that

  value = `some-command "@file"`

wouldn't work.  You would have to kludge

  backticcommand = '`some-command "' . file . '"`'
  value = @backticcommand

On the other hand, if you really need the "@" in the backtic
command, you would have to count the quotes.  This isn't really
intuitive.

That said, I think this patch is a small change which clearly
extends the power of gnuplot.


Juergen


PS: Just curious: Is there a reason that string_expand isn't called
from within scanner just like substitute?