Re: move constructors likely a requirement

[John W. Eaton]

On 9/5/19 2:57 AM, Carlo De Falco wrote:

> it seems this is the list of all the appearances of the pattern
>
>    T a = std::move (b);
>
> that were introduced by recent changes.
>
> attached are a patch that removes all warnings and the full log of make.

Thanks for the patch.  In the meantime, I guessed and made this change:

  http://hg.savannah.gnu.org/hgweb/octave/rev/d171d356767b

Sorry for the extra work.

> i tried to measure if the change makes any performance issue by timing 
> "make check"
> but I am encountering a segmentation fault that prevents that from working.
>
> strange enough, if I install then run __run_test_suite__ I see no crash and
> I get
>
>    PASS                            15684
>    FAIL                                5
>    REGRESSION                          5
>    XFAIL (reported bug)               38
>    SKIP (missing feature)             50
>    SKIP (run-time condition)          29

I build with PREFIX=/usr/local/octave/dev and sometimes have a copy 
installed there.  I just removed the /usr/local/octave/dev directory 
tree and tried "make check" and still don't see a crash.  My summary shows:

  PASS                            15748
  FAIL                                0
  XFAIL (reported bug)               25
  SKIP (missing feature)             37
  SKIP (run-time condition)          12

The missing features are all because the JIT compiler is disabled.  The 
other 12 tests that are skipped are all Windows-specific features.

jwe