Re: [Gcl-devel] Re: [Axiom-developer] list bug

[Camm Maguire]

Greetings!  

Have you considered something like:

(first (multiple-value-list (round (log 1000 10))))

3

?

Is this relevant to the 'duplicate member in set' bug we were
discussing earlier?  I'm referring to that reported by Juergen in:

=============================================================================
The last command in coercels.input converts the result
to a set. With gcl, the result contains duplicate 
elements. Problem does not occur with cmu cl.

Juergen Weiss

Juergen Weiss     | Universitaet Mainz, Zentrum fuer Datenverarbeitung,
address@hidden| 55099 Mainz, Tel: +49(6131)39-26361, FAX:
+49(6131)39-26407
=============================================================================

I still don't see where this is, but I'm still running a new build on
a fresh checkout to look for it.  To my understanding, this bug and
the hasCategory bug are the only two which appear to vary with the
underlying lisp.  And as you suspect a 'set' function in the latter as
well, they may be the same issue.  Just as a data point, I tried
loading a set-difference which reverses the order of the output to
test your hypothesis in interpsys before attempting to compile xpoly.
The compile still fails.  

BTW, please excuse my earlier hasty exuberance.  I knew something here
was recursive, and I knew axiom had to extend the stacks anyway to
some large value, so I just assumed the sequence would terminate if
there was enough room, and reported the source of the stack limitation
in my earlier email.  It does now appear to be an infinite loop.  And
while I need to verify, the infinite loop appears to lie completely
within recursively called knownInfo, being passed some erroneously
repetitive input.

I still would greatly appreciate the correct output from my earlier
posted modified knownInfo on a working system.  Also, this has
obviously worked with some gcl in the past.  Any details on the last
known working gcl setup?

Take care,

root <address@hidden> writes:

> Camm,
> 
> Bill Page pointed me to an email (which I either never received or
> it never hit my long-term memory) from Juergen Weiss that found the
> root of the display problem. The LOG10 function in GCL returns 
> rounded values that are used in the output length computation
> (src/interp/i-output.boot.pamphlet, line 843).
> 
> The GCL results are:
> log10(100) ==> 2.0
> log10(1000) ==> 2.9999999999999996
> 
> we take the floor of the results as the length of the number
> giving (floor (log10 1000)) => 2 rather than 3. Patching the
> generated boot.clisp code and reloading it fixes the problem.
> For now I'll add a fudge factor into the boot code.
> 
> Many thanks to Juergen and Bill.
> 
> Tim Daly
> address@hidden
> address@hidden
> 
> 
> _______________________________________________
> Gcl-devel mailing list
> address@hidden
> http://mail.gnu.org/mailman/listinfo/gcl-devel
> 
> 
> 

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