help-gnu-emacs
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: plists, alists, and hashtables


From: Stefan Monnier
Subject: Re: plists, alists, and hashtables
Date: Thu, 06 Aug 2015 15:12:54 -0400
User-agent: Gnus/5.13 (Gnus v5.13) Emacs/25.0.50 (gnu/linux)

> (alist-get 'a '((a) (b 1) (c . 2) d)) -> nil
> (alist-get 'b '((a) (b 1) (c . 2) d)) -> (1)
> (alist-get 'c '((a) (b 1) (c . 2) d)) -> 2

These are all consistent since your list is really

   ((a . nil) (b . (1)) (c . 2) d)

> (alist-get 'd '((a) (b 1) (c . 2) d)) -> nil ; no error

Just like Javascript, Elisp often errs on "don't signal an error" when
it bumps into something that's of the wrong type.  I don't like it very
much, but it has its advantages as well.

> The real map type in Emacs Lisp is the hashtable, I think.

I think that's a bias you get from other languages.

> For instance, the hashtable keys are unambiguously
> (hash-table-keys my-hashtable) and there's no looping or parsing.

For instance, the alist's keys are unambiguously (mapcar #'car my-alist) 
and there's no looping or parsing.

Also known as:

   For instance, the plist's keys are unambiguously (plist-keys my-plist)
   and there's no looping or parsing.

[ And no, please don't go and add plist-keys to subr-x.el.  ]

> However after perl and python (and awk) and... almost any language invented
> in the last 20 years, its rather backward.
> The efficiency is one thing

The efficiency argument is far from clear, since the cut-off point is
not too far from 100 last time I checked (IOW alist-get is faster than
gethash for tables of less than a hundred elements).  That was a while
ago, so it may be worth re-benchmarking it.  Also, I'd welcome patches
which optimize hash-tables for smaller sizes.

In any case, other than a vague complaint about Elisp being different
from your favorite language, I don't know precisely how to interpret
this thread.

More specifically, I highly doubt that a special «k1 v1 k2 v2 ...»
syntax for hash-tables would make much difference compared to
(hash-table k1 v1 k2 v2 ...) which you can get today with a very simple
`hash-table' macro.  I mean, you'd still have to say (gethash k m),
whereas you'd probably want something like m.k, etc...
For better or for worse, Elisp is not Python.


        Stefan




reply via email to

[Prev in Thread] Current Thread [Next in Thread]