On Mon, 2005-02-07 at 18:47 +0200, Martin Schlemmer [c] wrote:
> On Mon, 2005-02-07 at 09:46 -0500, Chris Gianelloni wrote:
> > On Mon, 2005-02-07 at 12:07 -0200, Fernando Serboncini wrote:
> > > I don't think I see the light here.
> > > What's the use of virtualx.eclass without X support? It basically uses
> > > xhost and Xvfb to create another DISPLAY, so I don't think it even runs
> > > without virtual/x11 installed.
> > > 
> > > And since you're talking about virtualx dependency... 
> > > 
> > > Last week I had a problem with virtualx eclass because it uses Xvfb
> > > which isn't installed if xorg-x11 is built with USE="minimal". If
> > > there's no other alternative, I think virtualx should "built_with_use
> > > xorg-x11 minimal" and fail nicely if this happen. 
> > > 
> > > Also, there should be a warning somewhere, so developers can know that
> > > virtualx only works on full xorg-x11.
> > 
> > Unfortunately, it is not that simple.  Packages should still be able to
> > compile if xorg-x11 was built with minimal.  Look at the recent changes
> > to pyopengl to see how it was done there.
> > 
> > I don't really see this one as being quite so easy to resolve.

I know what you mean. I was the one who fixed opengl. And the solution
was to remove "inherit virtualx", since there's no way to use it with
xorg-x11 minimal (Xvfb needed).

> One method might be to at least install Xvfb as well, even if
> USE=minimal?

It would be good to have Xvfb on xorg-x11 minimal. But I don't know if
it conflicts with USE="minimal" objectives.

[]s
Fernando

Attachment: signature.asc
Description: This is a digitally signed message part

Reply via email to