Strict Compile Mode

Ben Rubinstein benr_mc at
Tue Oct 9 04:48:00 EDT 2012

On 08/10/2012 21:15, Peter Haworth wrote:
> I was surprised to find that not all variables have to be defined in this
> mode.  It seems that using a variable in a repeat command such as "repeat
> with x=1 to 10" does not require x to be defined as a local variable, nor
> is an error flagged when x is used later in the handler, either within the
> loop or after it.  I can't decide if I like that or not.

> It also seems that sometimes LC constants have to beenclosed in quotes.
>   For example "Set the backgroundcolor of field "xyz" to empty" works fine
> but "set the backgroundcolor of field "xyz" to black" fails and I have to
> put quotes around black, even though black is colored in the script as if
> it is a known LC constant.  Not a big deal, just wondering why that is.

We thought those things were constants (as empty is); it turns out that they 
were really literals (although the dictionary describes them as keywords).   I 
think the script editor colouring is the same as for any other quoted string.

> Also, not really related to the rest of this, what does the Preference
> setting "Variable Preservation" do?  Can't find any reference to it in the
> User Guide.

It's effectively nullified by what I think is a bug

It should be what you set so that you're always in 'explicit vars' mode by 
default.  However, since the menu item "Variable checking" is sticky between 
launches, it is now as far as I know meaningless.

What would be really useful (as requested in above RQCC report) is to make the 
setting sticky per script (with the preference then affecting new scripts). 
This would enable one to be 'virtuous' by default, but not unduly 
inconvenienced when working with old or contributed code.  The problem I find 
is that if I have to fix eg a bug in a library, that's not 'explicit vars 
clean', I need to switch off the setting.  I may then forget, and lose the 
protection when I'm working on more modern scripts.

Sadly you can no longer vote for reports in RQCC...


More information about the use-livecode mailing list