NetNews Offline 2/NetNews Offline Volume 2.iso/news/comp/lang/c++-part2/10711

Internet Message Format | 1996-08-05 | 1.6 KB

Path: inforamp.net!ts10-07
From: [e-mail] (Randy Charles Morin)
Newsgroups: comp.lang.c++
Subject: Re: Question about visibility of local variables
Date: Sat, 09 Mar 96 18:00:36 GMT
Organization: MiddleWorld SoftWare
Message-ID: <4hsh0g$[e-mail]>
References: <4hgufv$[e-mail]> <4hifc6$[e-mail]> <4hj56k$[e-mail]> <[e-mail]>
NNTP-Posting-Host: ts10-07.tor.inforamp.net
X-Newsreader: News Xpress Version 1.0 Beta #4

In article <[e-mail]>,
   Mike Girou <[e-mail]> wrote:
>Well, maybe.  Certainly that was the way we did things in C.  But
>there has been quite a lot written about declaring/defining at first
>use, particularly with class variables.  These arguments include
>possible efficiency and the notion that variables needed to properly
>initialize may not be known when the "block" is entered.

I agree completely.  Sometimes, it is better to declare the variable later on
in a block.  But it is much more efficient to add blocks.

Example:

{
    BIGCHUNKOFMEMORY a,b,c;
    ...
    a=
    ...
    b=
    ...
    c=
    ...
}

..could also be ...

{
    ...
    BIGCHUNKOFMEMORY a=
    ...
    BIGCHUNKOFMEMORY b=
    ...
    BIGCHUNKOFMEMORY c=
    ...
}

..but this is also inefficient.

Instead of declaring stuff half way through a block,
it is better to add blocks...

{
    ...
    {
        BIGCHUNKOFMEMORY a=
        ...
    }
    {
        BIGCHUNKOFMEMORY b=
        ...
    }
    {
        BIGCHUNKOFMEMORY c=
        ...
    }
}

This is much more efficient then the previous and it is easier to read a block
since the variables will be declared at the beginning of the block.

Agrivar