Re: PATCH: function copy

[Philippe Troin <phil@xxxxxxxx>]

On Tue, 2019-07-16 at 09:56 +0100, Peter Stephenson wrote:
> On Mon, 2019-07-15 at 14:42 -0700, Bart Schaefer wrote:
> > On Mon, Jul 15, 2019 at 1:00 PM Peter Stephenson
> > <p.w.stephenson@xxxxxxxxxxxx> wrote:
> > >  
> > >  
> > > I've had this lying around for a while, wondering if there's more
> > > to it,
> > > but I can't think of it.
> > >  
> > > The point is that it's very easy internally to provide an
> > > interface to
> > > tweak standard functions to add arbitrary code before and after
> > > --- we
> > > have most of the support for this internally, and just lack the
> > > means to
> > > add a different name for a function, which this adds.
> > Emacs calls this "advice" and allows before/around/after variations
> > which can be added without having to redefine the existing
> > function.
> > I have a half-finished (that may be optimistic) module to provide
> > this
> > for ZLE widgets.  Handling the before/after is not too bad, but for
> > "around" you need a way to say "call the original function HERE"
> > which
> > you can then embed in another function that becomes the "around"
> > (and
> > which is called in place of the original everywhere except HERE).
> 
> That's basically what I showed in my example.
> 
> functions -c _std_fn _my_fn
> _std_fn() {
>   # do stuff here
>   _my_fn "$@"
>   # do stuff here
> }

I've been using the following hackish idioms:

   % foo() { print foo $@; }
   % () {
        local func=$functions[foo]
        eval "foo() { print before foo; { $func }; print after foo }"
     }
   % foo bar
   before foo
   foo bar
   after foo

   and

   % functions[foo_copy]=$functions[foo]

   Of course these cause reparsing of the function body, so "function -c"
   may still make sense.
   I like the idea of a full advising system.

   Phil.