Christof,
I had already downloaded your structure class, and I highly recommend it
for a good structure substitution. I know that had to take alot of work,
so thanks from everyone. I know it has saved me a few hours to say the
least.

>>1) Adding a THISCLASS access method for easier class programming

I often use container classes for pageframe programming and addobject
the class to the page when it is needed. The challenge is: Go to the
class designer and start a container class, add several objects with
their own containers. Now if you use CLASS properties and methods, which
is the right thing to do, you have to This.Parent.Parent.Parent.Parent.Parent
to get to them, since you can't use ThisForm because your coding in class.
And since fox can't pcode this syntax, class programming with containers can quickly bloat the exe

>>2) I like the structure class workaround, but for true API interfaces, I
need callback capability and addressof like VB or C. Even the clunky way we have to just get Hwnd, why not make it a property of the form? There was another suggestion of providing controls that are true windows subclasses, with the Hwnd and everything available. These controls could exist in addition to the standard bitmapped controls for backward compatibility. To help phase out the bitmapping of forms without breaking backward compatibility.
What do you think?


>>3) 'Filtered Result Sets' fixed to work with commands in the same manner

This item has really bitten alot of people. See the thread from hell 'Rushmore Design Flaw - HEADS UP!' for the complete skinny. I do agree that SQL server and views are the way to go, but there will still be a large free table developer base in our lifetime. So don't write it off just yet ;)



>Hi Ed,
>
>>1) Adding a THISCLASS access method for easier class programming
>
>Can't imagine what you mean with this.
>
>>2) Adding structure definitions for easier API programming
>> STRUCT...
>> cVar AS INTEGER
>> ENDSTRUCT
>
>Would

>DEFINE CLASS myStruct AS STRUCT
>   cVar = 0
>   cMembers = "i:cVar"
>ENDDEFINE

>work for you? In that case you can download my Struct class from the library here.
>
>>3) 'Filtered Result Sets' fixed to work with commands in the same manner that 'NOFILTER result sets' work. Especially record pointer commands and ERROR 1815 (sub-selecting a filtered result set error)
>> GO (rec)
>> RECNO()
>
>Just add NOFILTER if you want the same result, or create a filtered index. I can't see any sense in making an optimization technique slower... Beside that, the future definitely belongs to views in any variation and views are always created as physically existing result sets.
>
>
>Christof