[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: lisp style question
From: |
RG |
Subject: |
Re: lisp style question |
Date: |
Sun, 05 Dec 2010 21:45:59 -0800 |
User-agent: |
MT-NewsWatcher/3.5.1 (Intel Mac OS X) |
In article <87aakjkg2i.fsf@kuiper.lan.informatimago.com>,
"Pascal J. Bourguignon" <pjb@informatimago.com> wrote:
> RG <rNOSPAMon@flownet.com> writes:
>
> > In article
> > <6e2fe51c-ff4c-429f-b221-a3cbe23e958f@n2g2000pre.googlegroups.com>,
> > Katalin Sinkov <lispstylist@gmail.com> wrote:
> >
> >> how to conveniently costruct the list that goes with getf ?
> >
> > You can't construct anything with GETF. GETF is an accessor.
>
> This is not correct. GETF is special:
>
> CL-USER> (let ((plist '()))
> (setf (getf plist :k1) 1
> (getf plist :k2) 2)
> plist)
> (:K2 2 :K1 1)
>
> Setfers are able to do such things.
GETF is not unique in this regard. GETHASH works the same way. But
relying on GETF's ability to allocate storage for new keys is fraught
with peril, e.g.:
(defun make-new-plist () (list :key1 :value1))
(defun set-key (plist key value)
(setf (getf plist key) value))
(defun foo ()
(let* ((plist (make-new-plist)))
(set-key plist :key1 :new-value-1)
(set-key plist :key2 :value2)
plist))
(foo) ==> (:KEY1 :NEW-VALUE-1) ; What happened to key2?
I think it's better to advise beginners to avoid such things rather than
try to explain the sometimes subtle distinction between places and
first-class mutable data objects.
rg