1. Fw: A few more shrouding niggles
(resent - ISP was down)
----- Original Message -----
From: "Derek Parnell" <ddparnell at bigpond.com>
To: <EUforum at topica.com>
Subject: Re: A few more shrouding niggles
----- Original Message -----
From: "Robert Craig" <rds at RapidEuphoria.com>
To: "EUforum" <EUforum at topica.com>
Subject: Re: A few more shrouding niggles
>
>
> Pete Lomax wrote:
> > I had a few local variable names clashing with the global routines
> > value(), upper(), lower(), and string() [last one is in win32lib].
> >
> > The offending routines are bits_to_int() in machine.e (value),
> > reverse() (upper and lower), value() and wildcard_match() (string).
> >
> > It only causes a problem if routine_id is used with a variable
> > parameter in the program being shrouded.
> >
> > I've renamed the local variables/parameters in my local copies of
> > machine.e, get.e, and wildcard.e, to get round this.
> >
> > Of course I'd prefer to see the shrouder "forget" local variable names
> > at the end of each routine.
>
> Thanks for reporting this.
>
> In almost all cases, people hard-code the name
> of a particular routine into their routine_id() call.
> In the rare case where a variable or expression is passed
> to routine_id(), the shrouder must mark all routines that
> are visible at that point in the code, so that none of them
> can be renamed, just in case they are targetted by that routine id call.
> This tends to put naming pressure on the shrouder and
> this dynamic routine id case hasn't been tested as well.
> I intend to go through all of this very carefully
> before the next release.
And the next version of win32lib is a heavy user of expressions in routine_id.
