I wrote: > It is exhibited by all f"0 y where 0 e. $ y (whether the "0 is, to > borrow a phrase, > express or implied). Basically you're asking to operate on all the scalars > of an array > which contains no scalars.
For completeness, I guess I should note that 0 is not a special number here. All phrases of the form z=. f"r y where y contains no r-cells will produce empty results (i.e. 0 e.$z ) and hence empty displays. An analogous statement is true for phrases of the form z=. x f"(lr,rr) y . Again, all ranks here can be express or implied (the natural rank of a verb, or specificed explicitly using the rank conjunction). A quick adaptation of Roger's model of the rank conjunction at: http://www.jsoftware.com/jwiki/Essays/Rank#head-9912249d00fd010b059214f0e6cf59dbe3ce3801 Leads to the adverb qr : rk =: [EMAIL PROTECTED] er =: (0:>.(+rk))`(<.rk) @. (0<:[) fr =: [EMAIL PROTECTED] }. [EMAIL PROTECTED] qr =: 1 : 0 'mm ll rr'=. 3$&.|.m (0 e. mm fr ]) : ( (ll fr [) +.&(0&e.) rr e. ] ) ) which, given a rank as its argument, m , will produce a verb indicating whether f"m y or x f"m y will produce an empty result (ie a result without volume, ie an empty display) . The relevant call for Gosi's example is: 0 qr '' 1 Which indeed indicates 6!:0"0 '' will produce an empty display. In fact, the adverb can be extended to accepts verbs as arguments directly, rather than ranks: qv =: 1 : '(u b. 0) qr' 6!:0"0 qv '' 1 -Dan PS: Apologies if the adaptation is buggy, but I didn't want to post a cheating hack like 0 e. $@:(<"mm) . ---------------------------------------------------------------------- For information about J forums see http://www.jsoftware.com/forums.htm
