Forcing string comparisons, or When is "0" not 0?

[Mark Waddingham]

On 2015-10-14 22:19, Devin Asay wrote:
> Shouldn’t there be a way to force a string comparison? I know LC tries
> to be all helpful about casting numerals as numbers, but what if I
> want to know if it’s the exact string?

I think this is one of those things which has come up periodically over 
the years...

We added 'is really a' operators in LC8 to help with writing code which 
needs to preserve values exactly (the main use-case is lcVCS) - so we 
have been considering an 'is really' operator.

(It occurs to me this morning that perhaps these should be 'x really is 
a string', or 'x really is y' as opposed to 'x is really a string', or 
'x is really y' - I'm not sure which is 'more correct' in English)

The 'is really a' operators check the internal (dynamic type) of the 
value, by-passing any type-coercion:
   'x is really a string' -> returns true if the current value of x is 
(internally) a string
   'x is a string' -> returns true if x can be converted to a string

So, the 'is really' operator would do much the same thing:
   'x is really y' -> returns true if the internal types of x and y are 
the same, and they are the same value
   'x is y' -> if x and y can be converted to numbers then compare as 
numbers else compare as strings

The problem with 'is really' is that to truly understand what it is 
doing, you have to explain about whilst LiveCode is a 'typeless' 
language (assuming you ignore the existence of arrays ;)), the engine 
still has a notion of distinct types internally (it needs to store the 
values in memory in some chosen representation after all) and the 
internal type of a value depends on how the value was last produced:
     put "0" + 0 into tVar1 -- tVar is really a number
     put "0" & 0 into tVar2 -- tVar is really a string
     put tVar1 is really tVar2 -- false

The other option (which has the potential advantage of not exposing the 
7.0+ under-the-hood dynamically typed nature) is to have an explicit 
'compare as string' operator (for purposes of exposition let's call it 
is_string) for now. The action of such an operator would be to convert 
both sides to strings (if possible) and then compare:
     put "0" is_string "0." -- false
     put 0+0 is_string char 1 of "0." -- true

This is subtly different from is really:
     put 0 + 1 is really "1" -- false
     put 0 + 1 is_string "1" -- true

Indeed, if we imagined that we had 'as <type>' operators then:
     x is_string y <=> (x as string) is really (y as string)

So, anyway, a couple of potential solutions (I think 'is really' is a 
useful compliment to the 'is really a' operators, the question is 
whether there is a nice syntax for is_string and whether it is a useful 
thing to have).

Warmest Regards,

Mark.

> I guess I could do this dance:
> 
>   if char 1 of fld “display” is “0” and char 2 of fld “display” is NOT 
> “0” then…
> 
> It’s seems to complicated for such a simple thing, especially for
> explaining to novice programmers. Maybe I’m missing something obvious.
> 
> Devin
> 
> 
> Devin Asay
> Office of Digital Humanities
> Brigham Young University
> 
> _______________________________________________
> use-livecode mailing list
> use-livecode at lists.runrev.com
> Please visit this url to subscribe, unsubscribe and manage your
> subscription preferences:
> http://lists.runrev.com/mailman/listinfo/use-livecode

-- 
Mark Waddingham ~ mark at livecode.com ~ http://www.livecode.com/
LiveCode: Everyone can create apps