[Gcl-devel] (no subject)

[Matt Kaufmann]

Hi, Camm --

Thanks very much for the info about #@ and #!.

I'm afraid I don't have much of an idea about "real world" use for
fixnums.  In a local project, I think we take advantage of CCL's large
fixnum range on 64-bit machines:

  ? most-positive-fixnum
  1152921504606846975
  ? most-negative-fixnum
  -1152921504606846976
  ? (expt 2 60)
  1152921504606846976
  ? 

Of course, that's probably kind of irrelevant, since you were asking
about fast fixnums rather than fixnums.  I don't have much else to
offer; sorry.  I'll follow up with someone who may have more of a
sense about your question.

Thanks --
-- Matt
   Date: Mon, 22 Apr 2013 13:34:11 -0400
   From: Camm Maguire <address@hidden>

   From: Camm Maguire <address@hidden>
   To: Matt Kaufmann <address@hidden>
   Subject: Re: #@ and #! readers
   References: <address@hidden>
   cc: address@hidden
   Date: Mon, 22 Apr 2013 13:34:11 -0400
   In-Reply-To: <address@hidden> (Matt
           Kaufmann's message of "Fri, 19 Apr 2013 21:10:07 -0500")
   Message-ID: <address@hidden>
   User-Agent: Gnus/5.13 (Gnus v5.13) Emacs/23.4 (gnu/linux)
   MIME-Version: 1.0
   Content-Type: text/plain; charset=us-ascii

   Greetings!  Both of these appear to be unused, and set to the default
   dispatch macro which triggers an error.  I think they are leftovers from
   earlier attempts to read the .data file (appended to each compiled .o to
   initialize lisp data) using the lisp reader.  (see fasd.lisp, currently
   unused).  These are safe to reset AFAICT.

   On a separate note, regarding fixnum arithmetic and the 'immediate
   fixnums' supported in cvs head, I've been looking into this on my new
   64bit laptop.  This platform can easily support a ~ 500000000 fixnum
   table below the .text start *without any conversions* back and forth
   between machine representations.  In contrast, by default, an
   astronomically large table is set to the upper region of the 64bit
   address space which require shifts.  Do you have any sense on what a
   good fast fixnum table subset size might be in real word use?

   Take care,

   Matt Kaufmann <address@hidden> writes:

   > Hi, Camm --
   >
   > Readers for #@ and #! are defined in GCL 2.6.8pre, but we use them in
   > ACL2.  Do you see any reason why I can't redefine these readers for
   > ACL2?  Do you know what they do in GCL?
   >
   >   >(get-dispatch-macro-character #\# #\@)
   >
   >   #<compiled-function 000000000151b740>
   >
   >   >(get-dispatch-macro-character #\# #\!)
   >
   >   #<compiled-function 000000000151b740>
   >
   >   >
   >
   > Thanks --
   > -- Matt
   >
   >
   >
   >

   -- 
   Camm Maguire                                     address@hidden
   ==========================================================================
   "The earth is but one country, and mankind its citizens."  --  Baha'u'llah