Re: [Scwm-devs] Re: done testing key-mover-xtest.scm

[Greg B. <ba...@cs...>]

I agree with you that an interface to the "preferred" functionality is
in order, and that we need an interface to the primitive functionality,
as well.  I dislike global state side-effects, though, so I still prefer
an extra level of indirection that folks can use.  There may be other
reasons why folks want to send synthetic events (I can't think of
anything esp. compelling right now, but it's been a while for me).

Thanks,
Greg

Scott Lenser <sl...@cs...> writes:

> There are other ways to test this without exposing a dual interface though.
> For instance, you could expose a function that sets debugging flags that
> disallow the use of certain features of the X server.  Seems bad to make
> the client decide which one they want when we can easily determine it
> for them (perhaps on the scheme side, perhaps on the C side).
> 
> - Scott
> 
> > To test it, for one. :-)
> > 
> > Greg
> > 
> > Scott Lenser <sl...@cs...> writes:
> > 
> > > > Scott Lenser <sl...@cs...> writes:
> > > > 
> > > > > Ah, I see.  I didn't realize one was more robust than the other.  I didn't
> > > > > actually look to see how they were implemented much.  Couldn't this
> > > > > be folded into the move-pointer primitive so that you always get the
> > > > > best one that your version of X supports?  In any case, the function
> > > > > to move the virtual viewports is identical between the two versions.
> > > > 
> > > > It's possible, but sometimes you specifically want one or the other, so
> > > > the "smart" one would have to be another layer of abstraction.  The
> > > > stuff that's truly duplicated definitely should be factored out into
> > > > something separate.
> > > > 
> > > > Thanks,
> > > > Greg
> > > > 
> > > 
> > > When would you want the non-xtest one if it the xtest one is available?
> > > 
> > > - Scott