Re: Jiri's Widgets

new topic     » goto parent     » topic index » view thread      » older message » newer message

David Cuny wrote:

> Most windowing systems don't store the graphics in the window.
> Instead, they report to the window when a portion of the window has
> been 'damaged' (overlapped, off-screen, etc.) and the window's code
> is supposed to take care of it.

Too complex, David. I want to keep it as simple as practicable.

> Personally, I'm going to try double-buffering with my DOS window
> manager. When I draw into the window, I'll keep track of the window
> areas that were 'damaged', and only copy those portions to the
> screen when the drawing is finished.

(Just to show my total ignorance: people talk about double and even
triple buffering all the time. David, can you briefly explain what
those terms really mean. In simple language, please.)

Even your scheme sounds expensive. I do not really want to keep track
of every menu and message box in respect with every thing else that
happens to be on the screen. No wonder we need 600 MHz boxes with
Gigabytes of memory just to run some window systems. - I think, I will
simply not allow windows to slide out of screen.

> The windows look very nice, and have a look and feel all their own.
> Very slick.

I think you got it all mixed up, you really wanted to say 'very simple
minded'. If you consistently over do it like this with you kindness,
no one will believe a word of what you say soon ;). Thanks, David.



Earlier Everett L.(Rett) Williams wrote:

>> I do not redraw windows, I just save the rectangular image and
>> re-display that at the new location. Simpler and faster, I thought.
>> But, of course, off-screen pixels are undefined. Typically a bunch
of
>> zero, I suppose, and I finish with a black egg on my face, so to
>> speak. I am open to any reasonable suggestion...
>>
> Build you a virtual window and a displayable window. All logic
except
> actual transfer to write buffer happens to the virtual window.

I am not sure, Rett, I understand you completely. But if you mean
every thing happens first in the virtual plane and only the final
result is blasted on to the real screen, then I have a problem.
Firstly, to do it properly, I would have to go back to assembly, which
would be very stupid for a number of reasons I am reluctant to discuss
right now. I even gave up forth, and then C, to get some distance
between my brain and the metal buried under the grey plastic! And I am
sure you are not suggesting peeks and pokes, or are you? Secondly, it
is such a waste of effort to reblast the whole thing for every small
incremental change...

Thanks, guys. Good night. jiri

new topic     » goto parent     » topic index » view thread      » older message » newer message

Search



Quick Links

User menu

Not signed in.

Misc Menu