Re: removeprimes()

[Bill Allombert <Bill.Allombert@math.u-bordeaux.fr>]

>   Hi,
>
>   I noticed this interesting behavior.
>
>   ? addprimes(2*primes(64));length(addprimes)
>   64
>   ? removeprimes(addprimes);length(addprimes)
>   32
>   So removeprimes() cuts the size of the list of
>   declared primes in half.  I don't this this is how
>   removeprimes() is described to behave.
>
>   Thanks
>
>   Igor

The problem is that primetab is an universal object!

? addprimes(2*primes(64));length(addprimes)
%1 = 64
? x=addprimes;removeprimes(x);length(addprimes)
%2 = 0    

Bill.


Here a patch:
(I will update the CVS version this evening (Bordeaux time)).

===================================================================
RCS file: /home/megrez/cvsroot/pari/src/basemath/arith2.c,v
retrieving revision 1.2
diff -u -r1.2 arith2.c
--- src/basemath/arith2.c       1999/10/15 11:30:40     1.2
+++ src/basemath/arith2.c       1999/10/21 15:58:52
@@ -271,10 +271,20 @@
   long i,tx, av = avma;

   if (!prime) return primetab;
+  if (prime == primetab)/*remove all primes. B.A.*/
+  {
+    for (i=1; i < lg(primetab); i++)
+    {
+      gunclone((GEN)primetab[i]);
+      primetab[i]=0;
+    }
+    cleanprimetab();
+    return primetab;
+  }
   tx = typ(prime);
   if (is_vec_t(tx))
   {
-    for (i=1; i < lg(prime); i++) removeprimes((GEN) prime[i]);
+    for (i=1; i < lg(prime); i++) removeprimes((GEN) prime[i]);/*bug if prime==primetab*/
     return primetab;
   }
   if (tx != t_INT) err(typeer,"removeprimes");