Thanks Roger, When you get a minute, could you give some examples of why cell shape is better? My preference for the inner/outer shape description is that the inner shape is the component that describes the shape of the cell and the outer shape is the component that describes the shape of the frame. Together they make up the entire shape and the division between the two is the rank of the verb that is applied. There is a connection implied between the terms inner and outer that is not present between frame and cell (or outer and cell).
You have a much deeper understanding of the language and have seen complications that I have not. My ignorance of how cell shape is a better term is an opportunity to learn. Cheers, bob > On Jan 19, 2016, at 11:59 AM, Roger Hui <[email protected]> wrote: > > 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
