Sorry. Thanks for that correction. I quite like Robert's suggestion, though
I took it on board without adequate attribution. Wasn't trying to be
sneaky, though I see it might appear that way.

Pairing 'outer shape' and 'cell shape' does seem a bit unnatural. I think
we need a phrase like
(i) the argument's 'frame of cells', or
(ii) the argument's 'inner and outer shapes'.

The decision is not mine, but I can't see how adopting 'outer' on its own
is wise.

Here's an idea:

An argument has inner cells, each with shape 'inner shape'. Any argument
must always have one outer cell. It has shape 'outer shape'.

Any thoughts?
The alternative to frame / cell shape being discussed is outer shape / cell
shape, not outer shape / inner shape.  I don't believe anyone has used
"inner shape" before (and it's not as good as "cell shape").

On Tue, Jan 19, 2016 at 2:14 AM, Matthew Baulch <[email protected]>
wrote:

> Outer/inner makes perfect sense. Seems unlikely to lead anyone astray.
> To play devil's advocate, it might seem silly but maybe a newbie could
> guess that inner/outer shape relates to boxing. Is this paranoia? I don't
> know.
>
> The important question is: who is the terminology intended to serve? The
> answer is, of course, everyone. But in what proportions? There are
inherent
> trade-offs. As noted by Roger and others,
>
> Outer and inner shape
> Pros: fit together like a pair of gloves, suggest a dependency of some
sort
> (hopefully on rank!), is being adopted by Dyalog in a similar form (maybe
a
> pro?).
> Cons: has an unfortunate though slight suggestion of boxing.
>
> Frame:
> Pros: cells-in-frame concept makes some intuitive (and pictorial) sense,
> frame is (?) unused for terminology anywhere else in J so unlikely to be
> confused.
> Cons: has historical context around 'empty', tempts us to use the word
> empty, cells-in-frame concept breaks down when frame is empty (even though
> cells may still exist), no natural pairing with cells.
>
> Cells
> Pros: sort of makes sense...?
> Cons: has various meanings depending on the context, doesn't imply that
> shape or rank are at all important, no natural pairing with frame.
>
> I'm sure I've missed something. Anyway, I think there's a strong case for
> inner/outer shapes.
>
> On Tue, Jan 19, 2016 at 5:43 AM, Henry Rich <[email protected]> wrote:
>
> > I really like this suggestion.  "frame" makes sense for result: the
frame
> > is held fixed while the cell-results are coerced into the same shape,
and
> > then assembled using the frame.  For the arguments, "outer shape" shows
> the
> > dependence on the argument shape and (implicitly) the verb rank.
> >
> > I wonder whether we should try to move the documentation in this
> > direction.  There would need to be a general consensus in favor.
> >
> > Henry Rich
> >
> >
> > On 1/18/2016 11:52 AM, Roger Hui wrote:
> >
> >> The terminology originated in SHARP APL in the 1980s.  "Frame" was at
> >> times
> >> called "outer shape".  In some situations, "outer shape" may be a
> better,
> >> more easily understood term.  You know, cell shape and outer shape;
> outer
> >> shape is part of the shape; etc.
> >>
> >>
> >>
> >>
> >> On Mon, Jan 18, 2016 at 7:19 AM, Jose Mario Quintana <
> >> [email protected]> wrote:
> >>
> >> I would not be the one arguing for empty frame vs zero frame
terminology
> >>> :)
> >>>   (thanks for providing the context).
> >>>
> >>> Regarding frame, I meant it in the sense that Ken Chakahwata did: "to
> >>> have
> >>> a J definition of that fictitious primitive."
> >>>
> >>> Your executable model can, of course, readily address Ken's question
> and
> >>> other similar questions for specific instances (pointing out, albeit
> >>> rather
> >>> tacitly, that such J definition already existed, was my main reason
for
> >>> mentioning your article):
> >>>
> >>>     rk    =. #@$
> >>>     er    =. (0:>.(+rk))`(<.rk) @. (0:<:[)
> >>>     fr    =. -@er }. $@]
> >>>     cs    =. -@er {. $@]
> >>>
> >>>     (Y=. i.2 3 4)
> >>>   0  1  2  3
> >>>   4  5  6  7
> >>>   8  9 10 11
> >>>
> >>> 12 13 14 15
> >>> 16 17 18 19
> >>> 20 21 22 23
> >>>
> >>>     3 (er;fr;cs) Y    NB. effective rank; frame; cell shape
> >>> ┌─┬┬─────┐
> >>> │3││2 3 4│
> >>> └─┴┴─────┘
> >>>
> >>>     2 (er;fr;cs) Y    NB. effective rank; frame; cell shape
> >>> ┌─┬─┬───┐
> >>> │2│2│3 4│
> >>> └─┴─┴───┘
> >>>    _1 (er;fr;cs) Y    NB. effective rank; frame; cell shape
> >>> ┌─┬─┬───┐
> >>> │2│2│3 4│
> >>> └─┴─┴───┘
> >>>
> >>>
> >>>
> >>> On Sun, Jan 17, 2016 at 11:33 PM, Roger Hui <[email protected]
> >
> >>> wrote:
> >>>
> >>> I did not define them; Roland Pesch did: Empty Frames in SHARP APL
> >>>> <http://www.jsoftware.com/papers/EmptyFrames.htm>, 1986.  I did
> rename
> >>>> them
> >>>> to "zero frames".  Read the 1986 paper and you can decide for
yourself
> >>>> whether "empty frame" or "zero frame" is the better name.
> >>>>
> >>>>
> >>>>
> >>>> On Sun, Jan 17, 2016 at 5:28 PM, Jose Mario Quintana <
> >>>> [email protected]> wrote:
> >>>>
> >>>> The verb (frame) as well as the Zero Frame concept are defined in [0]
> >>>>>
> >>>> by
> >>>
> >>>> Roger.
> >>>>>
> >>>>> [0] Rank and Uniformity
> >>>>>      http://www.jsoftware.com/papers/rank.htm
> >>>>>
> >>>>> On Sun, Jan 17, 2016 at 7:11 PM, Ken Chakahwata <
> >>>>> [email protected]> wrote:
> >>>>>
> >>>>> My guess is that it would help if we could imagine that we had a
> >>>>>>
> >>>>> primitive
> >>>>>
> >>>>>> called 'frame' in the same way as we have one called 'shape' i.e. $
> >>>>>> Then one way to get to the precise meaning of frame is to have a J
> >>>>>> definition of that ficticious primitive. At a guess, this primitive
> >>>>>> requires the 'rank' of the cells in order to then return the
> >>>>>>
> >>>>> appropriate
> >>>>
> >>>>> frame.
> >>>>>> If we have an array of shape (x,y,z), and we stipulate cells of
rank
> >>>>>>
> >>>>> 3,
> >>>
> >>>> then the frame is presumably empty? Not sure of this... but anyhow,
> >>>>>>
> >>>>> just
> >>>>
> >>>>> a
> >>>>>
> >>>>>> thought...
> >>>>>>
> >>>>>> Enjoy
> >>>>>> ken
> >>>>>>
> >>>>>> -----Original Message-----
> >>>>>> From: Programming [mailto:[email protected]]
> >>>>>>
> >>>>> On
> >>>
> >>>> Behalf Of Henry Rich
> >>>>>> Sent: 17 January 2016 23:59
> >>>>>> To: [email protected]
> >>>>>> Subject: Re: [Jprogramming] Definition: Frame of an argument
> >>>>>>
> >>>>>> The terminology I use is an (x by y by z) array of cells, or an
> array
> >>>>>>
> >>>>> of
> >>>>
> >>>>> cells with frame (x,y,z), emphasizing that the frame is a (part of
> >>>>>>
> >>>>> the)
> >>>
> >>>> shape rather than an array.
> >>>>>>
> >>>>>> Henry Rich
> >>>>>>
> >>>>>> On 1/17/2016 6:16 PM, Raul Miller wrote:
> >>>>>>
> >>>>>>> Hmm... ok, reviewing
> >>>>>>> http://www.jsoftware.com/help/primer/frame_and_cell.htm 'frame'
> >>>>>>>
> >>>>>> does
> >>>
> >>>> get used that way.
> >>>>>>>
> >>>>>>> I was thinking of the frame as having a shape rather than being
the
> >>>>>>>
> >>>>>> shape.
> >>>>>>
> >>>>>>> Then again, since you can think of an array as being (for example)
> >>>>>>>
> >>>>>> an
> >>>
> >>>> (x,y,z) frame of cells, I do not think that my interpretation was
> >>>>>>> entirely incorrect, either. So I suppose I have gotten myself into
> >>>>>>>
> >>>>>> a
> >>>
> >>>> "much ado about nothing" sort of issue.
> >>>>>>>
> >>>>>>> Thanks,
> >>>>>>>
> >>>>>>>
> >>>>>>
> ----------------------------------------------------------------------
> >>>
> >>>> For information about J forums see
> >>>>>>
> >>>>> http://www.jsoftware.com/forums.htm
> >>>
> >>>>
> >>>>>>
> ----------------------------------------------------------------------
> >>>
> >>>> For information about J forums see
> >>>>>>
> >>>>> http://www.jsoftware.com/forums.htm
> >>>
> >>>>
----------------------------------------------------------------------
> >>>>> For information about J forums see
> http://www.jsoftware.com/forums.htm
> >>>>>
> >>>>
----------------------------------------------------------------------
> >>>> For information about J forums see
> http://www.jsoftware.com/forums.htm
> >>>>
> >>>>
----------------------------------------------------------------------
> >>> For information about J forums see http://www.jsoftware.com/forums.htm
> >>>
> >> ----------------------------------------------------------------------
> >> For information about J forums see http://www.jsoftware.com/forums.htm
> >>
> >
> > ----------------------------------------------------------------------
> > For information about J forums see http://www.jsoftware.com/forums.htm
> >
> ----------------------------------------------------------------------
> For information about J forums see http://www.jsoftware.com/forums.htm
>
----------------------------------------------------------------------
For information about J forums see http://www.jsoftware.com/forums.htm
----------------------------------------------------------------------
For information about J forums see http://www.jsoftware.com/forums.htm

Reply via email to