Re: znlog() behavior

[Bill Allombert <Bill.Allombert@math.u-bordeaux1.fr>]

On Wed, Jul 20, 2011 at 03:31:53PM +0200, Max Alekseyev wrote:
> And here is another bug, isn't it?
> 
> ? znlog(6,Mod(2,7),znorder(Mod(2,7)))
> %2 = 1

Sort of. (The code actually returns 3\2, see the attached patch which fix this issue).
The problem is that there is no interfaces for reporting "no solution".
For example:
? znlog(2,Mod(4,17),znorder(Mod(4,17)))
  ***   at top-level: znlog(2,Mod(4,17),zn
  ***                 ^--------------------
  *** znlog: gen_Shanks_log: supplied order (= 2) is incorrect.

but the supplied order is actually correct.
Maybe we should just do pari_err(consister,"gen_Shanks_log") instead.

Cheers,
Bill.

diff --git a/src/basemath/arith1.c b/src/basemath/arith1.c
index 6fc986d..01df34a 100644
--- a/src/basemath/arith1.c
+++ b/src/basemath/arith1.c
@@ -2104,6 +2104,7 @@ _Fp_easylog(void *E, GEN x, GEN g, GEN ord)
   {
     ord = dlog_get_ord(ord);
     if (!ord) ord = p1;
+    if (mpodd(ord)) pari_err(consister,"Fp_log");
     return gerepileupto(av, shifti(ord,-1));
   }
   avma = av; return NULL;