The point of my story about the rank-64 bug in IBM's APL of the 70s was
that hacking the system was the only use I ever encountered for
high-dimensional arrays.  I have never used anything higher than rank 6 in
actual code but that's over the past 50 years, so there's still time.

On Tue, Apr 6, 2021 at 10:30 AM Henry Rich <[email protected]> wrote:

> Space saving.  The V struct, describing a verb, fills 128 bytes, and I
> need to add to it.  If the maximum rank is reduced from 65535 to less
> than 256, I can avoid requiring a third cacheline.
>
> Henry Rich
>
> On 4/6/2021 12:20 AM, Roger Hui wrote:
> > I repeat my question, what are the benefits of restricting array rank to
> be
> > ≤ 63 ?
> >
> >
> >
> > On Mon, Apr 5, 2021 at 8:28 PM Henry Rich <[email protected]> wrote:
> >
> >> 'May be'.  Have you actually implemented it?
> >>
> >> I find it very hard to believe that a J sparse array would be the best
> >> choice for any database.  It seems to me to:
> >>
> >> * take lots of memory per value
> >> * have limited or inefficient queries
> >> * be very slow for updates, as the entire table must be copied
> >>
> >> Please prove me wrong.
> >>
> >> Henry Rich
> >>
> >> On 4/5/2021 10:04 PM, Igor Zhuravlov wrote:
> >>> On Sun, Apr 4, 2021, 11:52 PM Henry Rich <[email protected]> wrote:
> >>>> I propose that the maximum value for a rank, whether of a verb or a
> >>>> noun, will be 63.  Minimum verb rank will be _63.
> >>>>
> >>>> You got a problem with that?
> >>> A database in J may be implemented as a sparse boolean orthotope where
> >> each
> >>> axis corresponds to some attribute (field, column of relational table),
> >> and
> >>> value 1 ("true") marks tuple [1].
> >>>
> >>> A restriction proposed will limit a number of attributes to <64 for
> this
> >>> application. It would be sad but not fatal. I'd prefer to have this
> limit
> >>> larger, say, O(1e5).
> >>>
> >>> References:
> >>> [1]
> http://www.jsoftware.com/pipermail/general/2000-February/003102.html
> >>>       Jforum: Sparse Array as Database ?
> >>>       Roger Hui
> >>>       Thu Feb 10 19:06:50 HKT 2000
> >>>
> >>
> >> --
> >> This email has been checked for viruses by AVG.
> >> https://www.avg.com
> >>
> >> ----------------------------------------------------------------------
> >> For information about J forums see http://www.jsoftware.com/forums.htm
> >>
> > ----------------------------------------------------------------------
> > For information about J forums see http://www.jsoftware.com/forums.htm
>
>
> --
> This email has been checked for viruses by AVG.
> https://www.avg.com
>
> ----------------------------------------------------------------------
> For information about J forums see http://www.jsoftware.com/forums.htm
>


-- 

Devon McCormick, CFA

Quantitative Consultant
----------------------------------------------------------------------
For information about J forums see http://www.jsoftware.com/forums.htm

Reply via email to