🚀 go-pugleaf

Thanks for your input. But I can only consider this as my last solution
because I have to re-write too much (or not?)
I'm trying to solve it with Semaphore first (see my newer posting).


Phuoc Can HUA

"Oliver Rutsch" <nospam@sympatec.com> wrote in message
news:1046686601.962938@post.sympatec.net...
> Hi Phuoc ,
> [...]
> > execute() (of a TThread) where they are only called in sequence. Now I
have
> > the problem that sometimes (non-deterministic) the colors are wrong
(e.g.
> > reverse). Looks like the CopyMode is set wrongly. But I didn't have this
> > problem before. Unless I'm having the problem of concurrently access of
> > CopyMode. I've been debugging this for weeks :-(
> > May someone could help me.
> >
> > My question:
> > Does this has something to do with combination of CopyRect and TThread?
> > Are all the fuction calls in execute() running in the same thread, no
matter
> > where the function defined?
>
> I had similar problems using the canvas CopyRect functions in a thread.
> In my case it just stopped drawing, although the thread wasn't
> deadlocked. Later I found that the Handle property of the canvas got
> sometimes invalid, no idea why. Even locking the canvas didn't help in
> any way. Don't waste your time with VCL-GUI and multithreading. If you
> need it, use the win32 API functions (GetDC, ReleaseDC, BitBlt aso.), it
> worked very good for me.
> Another approach is to use the Synchronize function, but in this case
> the TTimer solution does the same and is easier.
> To answer your second question, no it doesn't matter where your code was
> defined. If it runs from execute it runs always in the thread.
>
> --
> Dipl. Ing. Oliver Rutsch
> EMail: orutsch at sympatec.com
> Sympatec GmbH Germany
>