[R] With which version of XFree86 can R compile?

[Prof Brian Ripley]

Almost all these issues are now resolved in the 1.9.0-beta sources, which 
I can now build against the XFree86 4.4.0 headers.  One that remains is

#define KeySym int

which is presumably there for some historical reason but should not be
with current X headers, and so needs a configure test.

On Sat, 27 Mar 2004, Prof Brian Ripley wrote:

> On 26 Mar 2004, Peter Dalgaard wrote:
> 
> > Prof Brian Ripley <ripley at stats.ox.ac.uk> writes:
> > 
> > > Note this is not R as distributed by R-core (which does not have
> > > dataentry_mb.c) but I believe some `Japanized' distribution.  People might
> > > want to check if the problem is in the changes (which I suspect) or R
> > > itself.
> > 
> > Yes. On the other hand, the original dataentry.c begins with
> > 
> > #include "Defn.h"
> > #include "Print.h"
> > 
> > /* don't use X11 function prototypes (which tend to ...): */
> > #define NeedFunctionPrototypes 0
> > #include <X11/X.h>
> > #include <X11/Xlib.h>
> > ...
> > 
> > and there is the off chance that the newer X11 is using a symbol in
> > our Defn.h or Print.h. If so, then we'd like to hear about it ASAP... 
> 
> I downloaded the XFree86-4.4.0 sources.  One issue is that 
> NeedFunctionPrototypes is no longer supported in constructs like
> 
> extern XFontStruct *XLoadQueryFont(
> #if NeedFunctionPrototypes
>     Display*            /* display */,
>     _Xconst char*       /* name */
> #endif
> );
> 
> which explains the warnings on incompatible types.  However, removing 
> 
> #define NeedFunctionPrototypes 0
> 
> does not cause the other errors on my XFree86 (8.3.3 I think).
> 
> More digging needed ....
> 
> Brian
> 
> 

-- 
Brian D. Ripley,                  ripley at stats.ox.ac.uk
Professor of Applied Statistics,  http://www.stats.ox.ac.uk/~ripley/
University of Oxford,             Tel:  +44 1865 272861 (self)
1 South Parks Road,                     +44 1865 272866 (PA)
Oxford OX1 3TG, UK                Fax:  +44 1865 272595