NetNews Offline 2/NetNews Offline Volume 2.iso/news/comp/lang/c-part2/12517

Internet Message Format | 1996-08-05 | 3.4 KB

Path: mail2news.demon.co.uk!genesis.demon.co.uk
From: Lawrence Kirby <[e-mail]>
Newsgroups: comp.lang.c
Subject: Re: Borland C's tmpnam()
Date: Mon, 01 Apr 96 12:41:02 GMT
Organization: none
Message-ID: <[e-mail]>
References: <[e-mail]>
Reply-To: [e-mail]
X-NNTP-Posting-Host: genesis.demon.co.uk
X-Newsreader: Demon Internet Simple News v1.27
X-Mail2News-Path: genesis.demon.co.uk

In article <[e-mail]>
           [e-mail] "Verne Arase" writes:

>In article <[e-mail]>,
>Lawrence Kirby <[e-mail]> wrote:
>
> >>I thought that only DOS/Intel offspring platforms were supported; that's
> >>why I limited my query to that specific compiler; else I would've
>included
> >>the target environment.
> >
> >Don't forget that you are posting to comp.lang.c where the bias is for
> >discussing the standard, platform-independent, C language. The place to
> >discuss the specific traits and extensions of Borland C is somewhere
> >like comp.os.msdos.programmer.
>
>Don't forget the subject line. I didn't ask a general question re:
>tmpnam(); I put it in the context of Borland C. If I'd asked the question
>in the context of CodeWarrior, I'd have expected someone with some
>Macintosh experience to respond.
>
>The newsgroup is called comp.lang.c, not comp.lang.x-platform.c.

Exactly - comp.lang.c is a platform-independent newsgroup. When you start
asking platform-specific questions you are posting to the wrong newsgroup.
However discussing what the standard requires is relevant to any compiler
that claims to be conforming.

> >>If you're stating that tmpnam() under Borland C does indeed check to see
> >>that the file name is unique, I'll quash that reluctance and change the
> >>default directory.
> >
> >I don't know what Borland C actually does. It it is a conforming
> >implementation it must ensure that the name it generates is valid and
> >doesn't already exist in the filesystem.
>
>Actually, it doesn't.
>
>All it does it insure that the file name doesn't exist in the current
>location in the file system. Which, I'd imagine, is what just about every
>implementation does.

In a hierarchical filesystem a file is uniquely identified by its full
path name. So to ensure "filename" doesn't already exist in the filesystem
it just nees to ensure that, for example,

    fopen("filename", mode)

is valid and doesn't refer to an existing file.

>It's be a lot more useful if you could specify a target directory, though
>the string that makes up that specification would be target environment
>dependant.

Unfortuately standard C has no concept of directories so this would make
no sense as a standard C feature. If I care about what directory it goes
in I usually care about what filename is generated and I'd generate my own.
A bigger problem is that filename generation and (possible) creation of
the file are distinct operations which could cause problems on a
multitasking system. What you really need is a function that will open/create
a new file, a bit like tmpfile() except that it also returns a file name.
You can't write this in standard C but on systems that support a POSIX-like
open() you can use that to build such a function.

--
-----------------------------------------
Lawrence Kirby | [e-mail]
Wilts, England | [e-mail]
-----------------------------------------