[Octave-bug-tracker] [bug #61300] integer range might exceed upper limit

[Arun Giridhar]

Follow-up Comment #13, bug #61300 (project octave):

Improved patch attached for the integer range bug in Range.h. This one gets
rid of the if-check and uses the standard technique of converting to double
and taking the floor. Since m_numel is of type octave_idx_type, I chose to
cast it to that type instead of type T.

All range tests from comment #12 pass with both integers and floating points.
Make check passes with no regressions and no new failures:

  PASS                            17028
  FAIL                                0
  XFAIL (reported bug)               27
  SKIP (missing feature)             43
  SKIP (run-time condition)          25


Please verify the behavior and the performance.

This is independent of the integer division bug in primes.m (patch in comment
#9).

(file #52361)
    _______________________________________________________

Additional Item Attachment:

File name: range_patch.patch              Size:1 KB
    <https://file.savannah.gnu.org/file/range_patch.patch?file_id=52361>



    _______________________________________________________

Reply to this item at:

  <https://savannah.gnu.org/bugs/?61300>

_______________________________________________
  Message sent via Savannah
  https://savannah.gnu.org/