Path: utzoo!utgpu!watmath!clyde!att!osu-cis!tut.cis.ohio-state.edu!mailrus!csd4.milw.wisc.edu!uxc!uxc.cso.uiuc.edu!m.cs.uiuc.edu!liberte
From: liberte@m.cs.uiuc.edu
Newsgroups: comp.emacs
Subject: Re: Summing Columns
Message-ID: <4300040@m.cs.uiuc.edu>
Date: 22 Mar 89 05:07:00 GMT
References: <5158@xenna.Encore.COM>
Lines: 25
Nf-ID: #R:xenna.Encore.COM:5158:m.cs.uiuc.edu:4300040:000:962
Nf-From: m.cs.uiuc.edu!liberte    Mar 21 23:07:00 1989


> /* Written  6:40 pm  Mar 20, 1989 by bard@THEORY.LCS.MIT.EDU in m.cs.uiuc.edu:comp.emacs */
>   (2) eta-reducing the '(lambda (x) (string-to-int x)) into
>       (function string-to-int).  (function is like quote, but 
>       possibly better for the compiler.)

Using (function string-to-int) causes the byte-compiler to generate
byte-code that dereferences the function cell of 'string-to-int. 
But since application of this function also causes dereferencing of 
the symbol - but in C code - I would expect that it is faster to
let the C code to it.  Therefore, (function string-to-int) would be
slightly *slower* than 'string-to-int in byte-code.

Where the function special form is useful is in things like:

   (function (lambda (x) (foo (bar x))))

This allows the byte-compiler to byte compile the lambda expression
since it is expected to be used as a function.


Dan LaLiberte
uiucdcs!liberte
liberte@cs.uiuc.edu
liberte%a.cs.uiuc.edu@uiucvmd.bitnet