Re: MacOSX: Octave.app 2.9.14

[David Bateman]

John W. Eaton wrote:
> [I'm moving this from the help list, since we are way beyond help
> now.  --jwe]
> 
> On  3-Oct-2007, Dmitri A. Sergatskov wrote:
> 
> | On 10/3/07, John W. Eaton <address@hidden> wrote:
> | 
> | > Is
> | >
> | >   sum (x .* x)
> | >
> | > really faster here than
> | >
> | >   sumsq (x)
> | >
> | > ?
> | 
> | No. sumsq is faster (at least on my machine).
> | But I am confused, isn't call to __vnorm__
> | better yet?
> 
> D'oh.  There is already a call to __vnorm__ in the current norm.m.
> The code we've been modifying only applies in the case of integer or
> sparse vectors.
> 
> Anyway, what about the following patch?  It makes norm a built-in
> function but calls an external __norm__.m function for cases it can't
> handle.  Currently the C++ code only handles what __vnorm__ did
> before.  But this removes the overhead of a .m file function for the
> cases handled by __vnorm__.
> 
> David, does this look OK to you?  With the patch, I see (for example):
>

That is much nicer, as the case that should be fast is :-)

D.