Re: goto: it's conceded
irv mullins wrote:
>
> David Cuny wrote:
> >
> >
> > I can't see the harm of adding GOTO to Euphoria. It adds real value for some
> >
> > people, and it doesn't (as far as I know) harm anyone by adding it.
> >
>
> Wow, it's good to hear [read?] from you once again!
>
> Anyway, "first, do no harm" is good advice not just for doctors.
> As far as I can tell, goto isn't going to bother me at all.
> There are already lots of keywords in Eu which I have never used.
> One more won't hurt. And I don't ever recall needing a variable
> named "goto":)
>
> Unlike "label" (dang it, EVERY graphic UI lib has something called a "label")
> Now that "label" is a keyword, I have to rewrite a bunch of code.
>
> Grumble, grumble....
You know, with some added smarts, the scanner could tell whether an identifier
can be used as a variable or might be a keyword. This won't work with a limited
set of keywords (return is the most obvious), but who would name a variable
"return"?
But for that purpose, it has to remember what kind the last token it sent was. I
never tried to benchmark this mod, brcause it was never asked for, and well,
until recenty, adding a new keyword was something bitterly fought by some, so
that the status quo always hld out.
When I get more available time, I'll check if this analysis is possible without
a noticeable performance penalty. If this works, perhaps it could be implemented
in 4.0 stable. alpha will be too early for me, as things go.
A last note: xin32lib has no problem, because it uses Labels, not labels ;)
CChris
|
Not Categorized, Please Help
|
|
