On Mon, 9 Jan 2006, Peter W. Draper wrote:
>> I think that configure should fail if the X11 header files are not found.
>
> For some packages, taken in isolation, that's probably correct, but for the
> system as a whole this would be problematic. Much of the data handling and
> processing can be done without X11 (think ADAM/NDF+80%KAPPA, some graphics
> works without X11 as well, postscript, Tek etc.), so it would be more
> legitimate to configure things to work with and without X11, as they can.
> Sounds like a large chunk of work.
>
I think that in most cases simply making the pgplot autoconf clever enough
to ignore X11 would result in most applications working. GKS could also be
patched to ignore X11.
The main problem is GWM since that is built as a dependency of both GKS
and GWM and requires X. So even if you built non-X PGPLOT/KAPPA (say,
using Aquaterm on Mac) you'd still have to work out how to sort out GWM
since you are not allowed to use configure to determine whether GWM is
present.
Note that tcl/tk will be problematic well before you get to the configure
part, or does the 'configure-deps' phase handle tk not being built?
I think all this rambling indicates that a well-placed die in configure
for GWM if X11 includes are missing will probably be early enough.
--
Tim Jenness
JAC software
http://www.jach.hawaii.edu/~timj
|