short name broken?

[Mark Waddingham]

On 2016-11-01 19:12, Richard Gaskin wrote:
> You're absolutely right.  When in doubt I check the
> Winkler-Kamins-DeVoto reference, and indeed they verify that all name
> queries will return a value, even when the property itself is empty.
> When using "the short name" on an unnamed object, WKD says it should
> return the ID form, as LC currently does.

I must confess I struggle to find any reasonable logic behind the 
current behavior of 'the short name' when the actual name of the object 
is empty. Indeed, I'd suggest that it is actually a bug (well, I guess 
'anomaly' by our current definition).

As 'the short name' is not meant to return a control reference, it is 
slightly unsettling that it does return a control reference in one case. 
Furthermore, one would hope that the following would always be true:

    the short name of (control (the short name of tControl)) is the short 
name of tControl

However, if tControl has an empty name, then this actually throws an 
error - breaking what would appear to be a perfectly valid invariant.

Of course, I guess the motivation here is that 'you shouldn't reference 
unnamed objects by name' (presumably because objects are generally 
unnamed until explicitly named, and thus you get a lots of objects with 
the empty name making it usually ambiguous). However, given that it 
probably just as frequent to get multiple objects with the same name 
(e.g. lots of buttons with name 'button'); I'm not sure the empty name 
should be considered specially.

Anyway, it is the way it has always been, as you rightly point out.

Warmest Regards,

Mark.

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