> From: Oleg Kobchenko
>
> The dictionary page numbering scheme is not random and
> is very thoughtful.
> One could come up with any numbers or
> names, such as ASCII scheme etc, which are pretty random.
> Instead what is done in the dictionary is nested grouping,
> which is numbered in three positions:
> - semantic group 0-6 (boolean, arithmetic, algebra etc)
> - operation order 0-4 (plus, times, minus, divide)
> - modifier 0-3 ({ {. {: {::)
>
> Letters are named with "dot" or "co" suffix.
>
> Makes perfect sense and platform/email friendly URLs.
> If you start thinking about it, you can soon even remember them.
Thanks for describing the rationale behind the scheme (and to Dan for pointing
out the helpndx script). The numbering scheme nevertheless "appears" to be
arbitrary (even to long-time users) and is certainly not newbie friendly.
IMO Dan's suggestion warrants more investigation. The question is whether we
can come up with a list of names for the root symbols that is more
expressive/easier to use/clearer than the current dictionary page names. If we
can't, then we can always reuse the current dictionary page names.
I've started a wiki page to help document/summarize the discussion:
http://www.jsoftware.com/jwiki/Vocabulary/Discussion
I've had a go at listing the symbols and some suggested names. It is tempting
to abbreviate some of them, but on balance I think unabbreviated symbol names
are clearer and will display better at the top of the wiki pages.
= equals
< lessthan/lt
> greaterthan/gt
_ underscore/uscore
+ plus
* star/asterisk
- minus
% percent
^ caret/hat
$ dollar
~ tilde
| verticalbar/vbar
. dot/stop/period
: colon
, comma
; semicolon/scolon
# hash/pound
! exclamationmark/emark/exclam
/ slash/fslash/forwardslash
\ backslash/bslash
[ leftbracket/lsqbr/leftsquarebracket
] rightbracket/rsqbr/rightsquarebracket
{ leftbrace/lclbr/leftcurlybracket
} rightbrace/rclbr/rightcurlybracket
" doublequote/quote/quot/dquot
` backtick/tick/btick
@ at/commercialat/asperand
& ampersand/amp
? questionmark/qmark
Other symbols used as punctuation in J:
( leftparenthesis/leftparen
) rightparenthesis/rightparen
' singlequote/squote/squot
> > From: Dan Bron <j...@bron.us>
> >
> > Ric wrote:
> > > What should we use for the names of the primitive pages
> >
> > This problem arose when the DoJ was first ported to HTML, and the
> solution
> > selected was to use a combination of opaque numeric codes
> > for graphical primitives, and (later?) abbreviated spellings for
> alphabetic
> > primitives. For example, = is d000, *: is d112, i. is
> > didot, and D: is D: ddcapco (and there are some extensions to these
> basic rules
> > to disambiguate, e.g. ^:n from ^:v). You can see
> > the full list with
> >
> > open '~system\extras\util\helpndx.ijs'
> >
> > Now, there might be some merit in maintaining this naming scheme, so
> that search
> > engines can infer a relationship between these
> > "new" vocab pages and the official vocab pages. But we can do that
> more
> > explicitly by linking from the new to the official (and
> > copying content, as Henry recommends), and in this era of SEO it
> might make more
> > sense to give descriptive names to the pages.
> >
> > The problem is, as you asked: what should those descriptive names be?
> I worry
> > about using symbolic names, which might cause
> > problems in browsers and anyway will lead to inconsistencies for
> those symbols
> > which can't be names (which you listed). But the
> > only other obvious option is using the (suggested) English name for
> each
> > primitive, but that too has problems: first, the English
> > names are merely suggestions, and other names could be used in
> addition or in
> > lieu, but a more serious problem is each primitive can
> > have multiple English names for its several roles (monad vs dyad or
> > noun-consuming operator vs verb-consuming operator), and worse
> > several primitives can share the same name (e.g. &. vs &.: ).
> >
> > So I suggest we come up with a page naming scheme which mirrors the
> spelling of
> > the primitive (like a modified and universal version
> > of the current alphabetic primitive naming scheme), and then set
> anchors in the
> > page for the suggested English terms. I think the
> > base term in a page name for any primitive should be the (possibly
> abbreviated)
> > English term for the basal glyph in the primitive
> > (as determined by typing that basal glyph into WP and seeing what
> page title
> > results).
> >
> > For example,
> >
> > * star
> > *. star. (I think this dot will be ok)
> > *: star:
> >
> > *. y star.#monad star.#square (note synonyms)
> > x *. y star.#dyad star.#nand
> >
> > ^: caret:
> > u^:v caret:#specified-power (or something)
> > u^:v caret:#calculated-power (ditto)
> >
> > We also might benefit by deviating from this scheme a bit (though we
> should
> > weigh these benefits against the cost of breaking
> > consistency), e.g. by rolling up all the closely-related composition
> functions
> > (@@: &&: &.&.:) into a single page, define their
> > common characteristics, and have a sub-heading for each flavor, which
> describes
> > how it differs from or contrasts with the other
> > flavors (and #REDIRECT the appropriately-named primitive page to this
> > subheading, e.g. amper.: --> under#infinite-under).
> >
> > I think having all these closely related functions on a single page
> would be
> > more helpful than separating them, as their
> > distinctions are a common stumbling block. We might also have a
> general page on
> > function composition that contrasts not just the
> > different flavors of each type of composition, but the different
> types of
> > composition themselves (@ vs & vs &.).
> >
> > Thoughts?
> >
> > -Dan
> >
> >
> > ---------------------------------------------------------------------
> -
> > 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
