On Donnerstag 18 März 2010, Wolfgang Rohdewald wrote:
> On Donnerstag 18 März 2010, Sune Vuorela wrote:
> > I don't mind cmake gently telling/warning that $stuff is
> > missing. but failing to build (or actually just not
> > installing it) is what I'm against.
> 
> that is how kajongg cmake now behaves.

I am afraid I have to re-open this thread.

Jonathan Riddell made this change to CMakeLists.txt,
commit 1131550:

> Give error messages if runtime depends aren't found, also
> note way to install without runtime depends

From this thread I got the impression that this is not
the best way - especially not for python programs.
Warnings should be given but not errors.

So I tried to contact Jonathan, I cited this thread as
reference.

That was three days ago, but I did not yet get an
answer from him. And I will be away for 3 weeks
starting tomorrow.

I am fine with either variant - I just think it might
be helpful to have clear guidelines about this.

-- 
Wolfgang