🚀 go-pugleaf

dspencer@mproj.retix.com (David Spencer) writes:

> I've never understood why the second arg to completing-read, 'TABLE',
> is an alist.  As far as I can tell, the cdr's of each element are
> not used, so why not just make it a list of strings?

I think i know the, or at least an, answer to this one.

Completing read takes a vector of symbols as well as an alist.  The
vector serves as the flat table format, handy for those times when
you're dealing with, eg, the elisp obarray.  Given a match in the
obarray, you obtain an associated value via a suitable symbol lookup
function (eg, symbol-{value,function}), or whatever custom relation
you prefer.

Alists are suited to those purposes where the value is encoded in the
table.  You get the value once you have the key via the assoc and assq
functions...

I've been messing with the completion functions, extending intrinsic
minibuffer completion for incremental, in-the-minibuffer candidate
displays.  The seemingly arbitrary completion functions wind up
covering all the bases pretty nicely.

Ken
klm@nist.gov, 301 975-3539