[DUG] Compnent creation question

[Neven MacEwan]

Paul

Worth a crack though, Bristol! you have a good memory (or I make an 
impression)
AFAIF they only mak(d)e cars and planes

Re the VCL, I remeber talking to someone years ago who told me it was 
orinally boosted from a
smalltalk library, can't recall who the original author was

Neven
> Neven wrote: 
>
>   
>> Thats why I though a Items.BeginUpdate would suppress events 
>> on the strings 
>> Usually as someon else posted, if this is a problem they may have a 
>> preceding event
>>     
>
> Yeah. I saw that. I haven't checked the VCL source but I suspect even
> calling BeginUpdate will likely still involve the Items TStrings wrapper
> trying to talk to the dead COMBO to ask it to not repaint as items
> messages are sent. Catch 22.
>
> The destructor is obviously too late to talk to Items on a Tcombo (at
> least, unless you hold your mouth right). That's the problem with leaky
> abstractions (i.e. wrappers) and mismatched lifetimes. The VCL can only
> hide so much behind it's skirt whilst still tring to leverage the
> Windows builtins. 
>
> Don't get me wrong. Borland did a pretty amazing job, but I sometimes
> wonder why they spent so much time cleverly wrapping the nasty windows
> builtins and replicating the faulty dialog manager design.
>
> I suppose it all looked like a good idea when they started. i.e. first
> 90% done, only remaining 90% left to go.
>
> But using TCombo replacements written in pure Delphi would have been so
> much simpler as a design model.  It sure would have obviated the
> Kylix/CLX deadend for Borland, but anyway, that boat has already sailed.
>
> Or you can go and build your own boat :-) Did Bristol ever get into
> shipbuilding? <Insert analogy here>
>
> TTFN,
>   Paul.
>
> _______________________________________________
> NZ Borland Developers Group - Delphi mailing list
> Post: delphi at delphi.org.nz
> Admin: http://delphi.org.nz/mailman/listinfo/delphi
> Unsubscribe: send an email to delphi-request at delphi.org.nz with Subject: unsubscribe
>
>
>