Index: [thread] [date] [subject] [author]
  From: MenTaLguY <mentalg@geocities.com>
  To  : ggi-develop@eskimo.com
  Date: Fri, 24 Jul 1998 20:25:10 -0400 (EDT)

Re: required libs

On Fri, 24 Jul 1998, Andrew Apted wrote:

> MenTaLguY writes:
> 
> >  Which reminds me ... I need to prepare a
> >  new version of the terminfo target that doesn't try to simulate a
> >  framebuffer -- the current implementation is horrendously inefficient as a
> >  result of trying to do so.  (something like steady >25% CPU utilization on
> >  my otherwise quiescent p100 during the texttest demo -- bleah) 
> 
> Take a look at the trueemu & monotext code, especially UPDATE_MOD()
> defined in the .h files which is used to keep track of the dirty region.
> Only the dirty region needs to be transferred to the parent visual at
> ggiFlush() time.

Sure, I could do dirty region tracking.  Only problem is that if anyone
accesses the framebuffer directly, it'll get out of sync, because there is
no way to keep tabs on what people do to the emulated framebuffer directly. 
Only way to do reliable dirty region tracking is if you don't emulate a
framebuffer.  I'm just going to stop trying to emulate a framebuffer, and I
can actually just let ncurses' existing dirty region tracking code do its
thing for the present.

> >  > 		(dl) required
> >  
> >  Only required if dynamic linking is enabled -- man, we need to figure out
> >  how we want to deal with static systems like DOS; I don't think we've
> >  actually got a specific setup planned for that yet, do we?
> 
> No we don't.  LibGGI relies heavily on dynamic linking, and trying to
> statically link it seems like a nightmare to me.  Apart from any
> duplicated symbol problems (e.g. every sublibrary defines GGIdlinit...),
> there'd also need to some infrastructure to convert "load generic-blah"
> to "find & use the statically linked generic-blah"...  Maybe I am being
> too pessimistic ?

Well, IIRC Andy wants that DOS port ... we need to figure out SOMETHING...

It might be an interesting excercise cobbling together a dynamic linker for
DJGPP -- I expect it's doable.

-=MenTaLguY=-

Index: [thread] [date] [subject] [author]