AMBER Archive (2006)

Subject: Re: AMBER: Installation of Amber8 on Fedora Core 5. .

From: David A. Case (case_at_scripps.edu)
Date: Fri Jun 02 2006 - 14:27:59 CDT

On Thu, Jun 01, 2006, Don.Bashford_at_stjude.org wrote:
>
> Having said all that, your particular compiler errors look to me like
> a semi-bug in the LEaP code. ShellP.h, etc. are supposed to be
> *private* header files within the X11 implementation. User code is
> just supposed to use Shell.h (without the "P"). I wonder if the code
> would compile if you simply changed the include file names.
>

Thanks for pointing this out; it's a piece of X11 I was not familiar with.
Unfortunately, it does not work to just change ShellP.h to Shell.h, etc. in
our existing code base.

I can only say that this code was written two decades ago, much of the X part
by non-Amber people (trying to rely as little as possible on any X extensions,
and going back to relase 3 of X, as I remember.) In spite of problems like
the ones pointed out above, the code continues to work (as far as we know)
on all existing X11 implementations.

I haven't seen any X-implementation that has Shell.h but not ShellP.h, and
I suspect that the original poster simply did not have the proper X11
development libraries installed (or that they were not being found).

We believe that we *have* finally all of the 64-bit problems (at least enough
that we don't know of any problems with xleap on 64-bit machines). These
changes are available in Amber 9 and as bugfix.60 for Amber 8. But Amber 8
still requires some hand-tweaking of a configuration file to find the correct
X11 libraries on most 64-bit OSes. For people who can get access to both
32-bit and 64-bit machines, the easiest path (for Amber 8) is often to just
create a 32-bit executable, and copy it over to the 64-bit machines (as
several recent posts have pointed out.)

Finally, I will say that LEaP is being completely rewritten (by Wei Zhang)
in a more modern idiom. This, of course, will probably have its own
portability problems :-( But it also means that, unless we have definite
evidence that calling ShellP directly is really the problem, the LEaP sources
probably won't get changed.

...regards...dac

-----------------------------------------------------------------------
The AMBER Mail Reflector
To post, send mail to amber_at_scripps.edu
To unsubscribe, send "unsubscribe amber" to majordomo_at_scripps.edu