NetNews Usenet Archive 1992 #19/NN_1992_19.iso/spool/comp/lang/c/12981

Internet Message Format | 1992-08-30 | 2.2 KB

Path: sparky!uunet!mcsun!news.funet.fi!hydra!klaava!wirzeniu
From: [e-mail] (Lars Wirzenius)
Newsgroups: comp.lang.c
Subject: Re: Novice Questions
Message-ID: <[e-mail]>
Date: 30 Aug 92 10:11:21 GMT
References: <[e-mail]>
Organization: University of Helsinki
Lines: 57

[e-mail] (Norman Hamer) writes:
> 1. Given a char array, say char foo[256];, is *foo the same as foo[0]
>and is foo+x the same as foo[x] in all implementations?

*foo and foo[0] are identical, and *(foo+x) and foo[x] are identical.
The latter is by definition: the language is defined that way.  The
former follows directly from the latter: foo[0] <=> *(foo+0) <=> *(foo)
<=> *foo.

> 2. What's the best way to read in a text configuration file? I've seen
>code like this:

> fgets(line, 80, fd);
> while(!feof(fd))
> {
>   /* Muck with the data */
>   fgets(line, 80, fd);
> }

Works, but is ugly, since it duplicates the read statement, and makes it
easy to miss one of them when changing the other one.  Doesn't check for
ferror.

> while(fgets(line, 80, fd) != NULL)
> {
>   /*Muck with the data */
> }

Preferred, and the way that practically everybody does it.  An idiom in
C.  Concentrates all the line reading logic into one place and has only
the data mucking inside the loop, and nothing else there, so the data
mucking is nicely isolated.

> while(!feof(fd))
> {
>    fgets(line, 80, fd);
>    if(feof(fd))
>        break;
>    /* Muck with the data */
> }

Even uglier than the first one, although more similar to Pascal (that is
not a virtue, although not necessarily a bad thing either).  Includes an
unnecessary goto (spelled break here), which confuses the logic of the
loop.  Also, this doesn't check for errors, you'd need to check for
ferror as well, but then it gets uglier and uglier.

> Which one is best and why?  Which is the least likely to provide
>erroneous results (the last one if you forget the second feof() will
>sometimes leave you with a portion of the last line... not all of it and
>not a null string)

The second one is best, 'cause it's the simplest and most common way of
doing it, and the most difficult to get wrong.

--
[e-mail]