#17526: Bitset doctest failures on OS X
-------------------------------------+-------------------------------------
Reporter: jdemeyer | Owner:
Type: defect | Status: needs_review
Priority: blocker | Milestone: sage-6.5
Component: misc | Resolution:
Keywords: | Merged in:
Authors: Jeroen Demeyer | Reviewers:
Report Upstream: N/A | Work issues:
Branch: | Commit:
u/jdemeyer/ticket/17526 | 0d1e1497735026f12f1cb06c5ef919f462d0bd23
Dependencies: | Stopgaps:
-------------------------------------+-------------------------------------
Comment (by SimonKing):
Let me try to put some semantic on the existing code (i.e., without your
patch).
We shift by `n` bits. The variable `nlimbs` denotes the number of limbs
that are fully discarded by the shift.
Expl: If a limb comprises 32 bit and we shift by 37 bits, then `nlimbs` is
1.
The variable `shifted_limbs` denotes the number of limbs of `a` that carry
data to-be-put into `r` (provided that `r` is large enough).
Expl: If `a` comprises 5 limbs in the example above, then we have
`shifted_limbs=4`.
The variable `nbits` is the remainder of `n` by the limb size, hence,
`nbits=5` in our example.
If `r` is large enough to contain `shifted_limbs` limbs, then we can just
shift (if `nbits>0`) resp. copy (if `nbits==0`). Otherwise, we can only
shift `r.limbs` limbs.
If in our example `r.limbs==3<shifted_limbs==4`, then we would like to
omit limb number 0 of `a`, and then take limbs 1, 2, 3 of `a`, shift it by
5 bits, and put the result into limbs 0,1 and 2 of `r`. In addition, we
want that the first 5 bits of limb 4 of `a` shall appear as the 5 highest
bits of limb number 2 of `r`.
However, if `r.limbs>=4==shifted_limbs`, then it is enough to shift limbs
1,2,3,4 of `a`, shift by 5 bits, and put the result into the limbs 0,1,2,3
of `r`. There are no additional 5 bits to be put somewhere. However, it is
needed to clear the top bits of `r`: It could be that `r` does not use all
bits of its limb number 3, and hence there might be some bits obtained
from `a` that should be discarded.
Hence, I believe the following solution would be easier:
{{{
#!python
if shifted_limbs <= r.limbs: # here I changed "<" to "<="
if nbits:
mpn_rshift(r.bits, a.bits + nlimbs, shifted_limbs, nbits)
else:
mpn_copyi(r.bits, a.bits + nlimbs, shifted_limbs)
# Clear top limbs (note that r.limbs - shifted_limbs >= 1)
mpn_zero(r.bits + (r.limbs - nlimbs), r.limbs - shifted_limbs)
else:
# Number of limbs to shift is r.limbs
if nbits:
mpn_rshift(r.bits, a.bits + nlimbs, r.limbs, nbits)
# Add the additional bits from top limb of a
r.bits[r.limbs-1] |= a.bits[r.limbs+nlimbs] << (GMP_LIMB_BITS
- nbits)
else:
mpn_copyi(r.bits, a.bits + nlimbs, r.limbs)
# Clear bits outside bitset in top limb
bitset_fix(r)
}}}
--
Ticket URL: <http://trac.sagemath.org/ticket/17526#comment:5>
Sage <http://www.sagemath.org>
Sage: Creating a Viable Open Source Alternative to Magma, Maple, Mathematica,
and MATLAB
--
You received this message because you are subscribed to the Google Groups
"sage-trac" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to [email protected].
To post to this group, send email to [email protected].
Visit this group at http://groups.google.com/group/sage-trac.
For more options, visit https://groups.google.com/d/optout.
