conversions between vectors

[Joseph S. Myers]

On 22 Dec 2001, Aldy Hernandez wrote:

> we have no support for any kind of operations on vectors.  dan berlin
> had a patch for + and - a few weeks ago, but no one approved AFAICT.

We can't expect the intermediate states of work on vector support to be
particularly useful.  The final state will allow + when valarrays allow
it.

> from what i've seen in vector extensions, signed and unsigned additions
> for vector types are the same.  i don't think there's overflow.

By default one would suppose that + on vectors has the elementwise effect,
with undefined behaviour on overflow if the operation on individual
elements is undefined.  I doubt that we should change this, even if the
machine instructions define things more precisely.

> > The programmers should normally be using the much less cumbersome +
> 
> i absolutely agree.  so, let's get Dan's patch approved.

Does Dan's patch include docs?  Let's first get in docs for vectors
corresponding to valarrays (effectively explaining those parts of
valarrays for C programmers), stating what is and is not currently
supported, then start adding the semantics (updating the docs as each part
is implemented).  But unless you can produce a bug that this fixes, this
ought to be 3.2 (or some new branch) material rather than for the mainline
now.

> however, what do you do for a function:
> 
> 	void foo (vector int, vector int);
> 
> that operates on vector int's but has the same effect for vector
> unsigned as for vector signed?  are you going to explicitly require a
> cast?  that's the part i don't like.

Define two such functions?  Define an alias for the function?  Suggest to
the programmer that it might be safer and clearer if they don't use both
signed and unsigned together?

-- 
Joseph S. Myers
jsm28@cam.ac.uk