sympow seems to have presented some problems on Solaris 10 on x86. I've tried
looking at this source code, and it makes little sense whatsoever to me.
William has tried to clean up the 'Configure' script, though it's still very
messy, with things like
SH=`whichexe sh` && echo "#define SH \"$SH\"" >> $CONFIG
if [ -z "$SH" ]; then
echo "**ERROR**: Could not find sh"; exit 1;
else
echo "SH = $SH"
fi
Anyway, that aside, lets get to the C source code.
I tried to compile this with the Sun compiler, thinking that would help trace
the problem. Well, it indicates a problem on line 68 of compute2.c
/bin/cc -O3 -c -o compute2.o compute2.c
"compute2.c", line 68: void function cannot return value
Looking at line a dozen or so lines either side of line 68, in the function
pth_coeff_from_ap, we see:
void pth_coeff_from_ap(int W,llint p,int ap,int sp,QD x) /* p<2^50 */
{QD t={0.0,0.0,0.0,0.0},u={0.0,0.0,0.0,0.0},v={0.0,0.0,0.0,0.0}; int i;
QD y={0.0,0.0,0.0,0.0},z={0.0,0.0,0.0,0.0}; int W2=2,W3=3,W4=4; QD ppow[16];
if (W<4) {W4=W; if (W<3) {W3=W; if (W<2) W2=W;}}
QD_copy(wmax,QD_zero,x); if ((HECKE) && (ap==0)) return;
if (HECKE) return power_trace_from_ap(W,p,ap,sp,x); /* LINE 68. !!!!!!! */
if (ap==0) {if (!(sp&1)) {t[0]=(double) (-p); QD_powi(W,t,sp/2,x);} return;}
switch(sp)
{case 1: x[0]=(double) ap; return;
case 2: x[0]=(double) ap; x[0]=x[0]*x[0]-(double) p; return; /* a^2-p */
case 3: y[0]=(double) ap; x[0]=y[0]*y[0]-(double) (2*p);
QD_mul1(W2,x,y[0],x); return; /* a^3-2pa */
Se we can see that if(HECKE) this is supposed to return a value from
power_trace_from_ap
But looking at power_trace_from_ap, we see it's declared as returning noting
(void)
void power_trace_from_ap(int W,llint p,int ap,int sp,QD x) /* p<2^50 */
{QD t={0.0,0.0,0.0,0.0},u={0.0,0.0,0.0,0.0},v={0.0,0.0,0.0,0.0}; int i;
QD y={0.0,0.0,0.0,0.0},z={0.0,0.0,0.0,0.0}; int W2=2,W3=3,W4=4; QD ppow[16];
if (DEBUG) printf("power_trace_from_ap %lli %i %i\n",p,ap,sp);
if (W<4) {W4=W; if (W<3) {W3=W; if (W<2) W2=W;}}
QD_copy(wmax,QD_zero,x); if (sp==1) {x[0]=(double) ap; return;}
y[0]=(double) ap; y[0]=y[0]*y[0];
QD_copy(W4,QD_one,ppow[0]); QD_copy(W4,QD_zero,ppow[1]);
I don't know how gcc manages to compile this myself, but when it does not work,
its hard to know what one is supposed to do when the code just looks wrong to me.
Likewise pth_coeff_from_ap is declared as void, but returns a value.
To me, this is just buggy gcc compiling code it should never compile. The fact
the formatting of the code is a bit non-C like probably confuses gcc a bit.
Anyway, does anyone know how to sort this out, as I don't have a clue. It's
simply not valid C, and I have no idea what the intension is either.
Dave
--
To post to this group, send an email to sage-devel@googlegroups.com
To unsubscribe from this group, send an email to
sage-devel+unsubscr...@googlegroups.com
For more options, visit this group at http://groups.google.com/group/sage-devel
URL: http://www.sagemath.org
