Hi Pritpal and all, On 2010 Jan 30, at 23:28, Pritpal Bedi wrote:
> > Hello Everybody > > I am in the middle of overhaul of Events and Slots protocols > which is aimed at to get rid of the GC pointer reference counting. > > I have changed the parameters how Events and Slots are calling > .prg level code block > s. Current implementation is dispatching codeblock > as ( prg equivalent ) > > Eval( EventOrSlotBlock, pObject, ... ) > > The first argument is pointer to the object calling the codeblock, > and this is a GC collectible pointer. And hence the current situation > renders it GPF prone. > > I have changed this arument to numeric zero instead of pointer to th object > just to retain compatibility with th existing code where codeblocks are > expecting first parameter as "something" which in reality has not been > used either of the implementation, viz., hbXBP and hbIDE. > > Changing argument from pointer to integer is working fine. > > But I am convinced that avoiding this parameter is best for the long run. > First for the simplicity of the prg code and also the speed. > > I know many of you have based your libraries on the above foundation, > and this change will require you to look deeply into your code, though > I think it will be fairly easy, so I propose to eliminate it altogether. > > Waiting for the responses for next few hours and then I will decide > in its favour. I don't have HBQT based programs, though I think we're so early in HBQT development in Harbour that it's simply natural to have changes like this until things settle to their ideal, final form. I believe most, if not all HBQT users are all closely following this list, and I'm sure everyone will adapt, especially if such change is required to have a solid base system. IOW it would be shooting ourselves in the foot, if we gave compatibility a priority over fixing things, at this early point of development. my 2 cents. Brgds, Viktor _______________________________________________ Harbour mailing list (attachment size limit: 40KB) [email protected] http://lists.harbour-project.org/mailman/listinfo/harbour
