[developers] PET 0.99.12 + MRS Build Errors

Stephan Oepen oe at csli.Stanford.EDU
Tue Apr 25 11:41:57 CEST 2006


> Francis' suggestion works. Apparently, we still need to push the
> architecture type onto the stack when configuring ECL, otherwise it
> won't be able to find its binaries.
> I have written a patch that implements this change.

this cannot be the ultimate solution, in my view.  i believe we have
observed in the past that ECL varies quite a bit in what features are
available on which architectures, but always forcing the architecture
to :i686 will likely be a road-block to a 64-bit version of cheap.  we
already attempt to cover a range of garden varieties in `ecl-patches',
but apparently a few more will be needed:

  (defvar %system-binaries%
    #+(and (or :i386 :i686 :pentium3 :pentium4 :athlon) :unix) "linux.x86.32"
    #+(and :x86_64 :unix) "linux.x86.64"
    #+(and (or :i386 :i686 :pentium3 :pentium4 :athlon) (not :unix)) "windows"
    #-(or (or :i386 :i686 :pentium3 :pentium4 :athlon) 
          (and (not (or :i386 :i686 :pentium3 :pentium4 :athlon)) :unix))
    (error "~&loadup: unable to determine system type; see file ~

eric and zhengbo, if you just bring up `ecl' interactively and do

  (pprint *features*)

then what do you get?

                                                       cheers  -  oe

+++ Universitetet i Oslo (IFI); Boks 1080 Blindern; 0316 Oslo; (+47) 2285 7989
+++     CSLI Stanford; Ventura Hall; Stanford, CA 94305; (+1 650) 723 0515
+++       --- oe at csli.stanford.edu; oe at ifi.uio.no; stephan at oepen.net ---

More information about the developers mailing list