NetNews Usenet Archive 1992 #30/NN_1992_30.iso/spool/comp/lang/cplus/18097

Internet Message Format | 1992-12-16 | 1.8 KB

Path: sparky!uunet!cs.utexas.edu!sun-barr!olivea!spool.mu.edu!umn.edu!csus.edu!netcom.com!erc
From: [e-mail] (Eric Smith)
Newsgroups: comp.lang.c++
Subject: Re: Why not nested functions ?
Keywords: nested functions, c++
Message-ID: <[e-mail]>
Date: 16 Dec 92 23:23:09 GMT
References: <[e-mail]>
Distribution: comp
Organization: Netcom - Online Communication Services (408 241-9760 guest)
Lines: 26

In article <[e-mail]> [e-mail] writes:
>My point is that you can *simulate* nested functions using
>a local class. To call the function you call a member function
>of the local class, you explicitely give the syntaxic context
>you need for the function copying the automatic variables
>you want to use in local class members variables.
>
>In one word : I *do* it, so it can be done.


The most important feature of a nested function is that you can call
it from an unrelated function, which has no knowledge of its context.

I don't see how this can be simulated without adding it as a feature
to the language, but I don't see any real reason for not adding that
feature.  It sure seems like it would make C++ a much more powerful
language, and that a lot of the most complicated code would be neater
and easier to work with.  It also doesn't seem very hard to implement,
using pointers to stack frame descriptors or something like that.

Or is there a way to simulate this feature using objects?  If so, can
someone please explain it clearly or refer us to a clear explanation?
The original poster seems to be describing a way to simulate a very
limited subset of this feature, in which it is necessary for all
functions invoking the nested function to have access to its context,
which sacrifices the biggest advantage of nested functions.