This is the mail archive of the
gdb-patches@sourceware.org
mailing list for the GDB project.
Re: [rfc v2][4/6] Readlink as file I/O target operation
- From: Eli Zaretskii <eliz at gnu dot org>
- To: "Ulrich Weigand" <uweigand at de dot ibm dot com>
- Cc: gdb-patches at sourceware dot org
- Date: Mon, 16 Jan 2012 08:29:33 -0500
- Subject: Re: [rfc v2][4/6] Readlink as file I/O target operation
- References: <201201161232.q0GCWN33024751@d06av02.portsmouth.uk.ibm.com>
- Reply-to: Eli Zaretskii <eliz at gnu dot org>
> Date: Mon, 16 Jan 2012 13:32:23 +0100 (CET)
> From: "Ulrich Weigand" <uweigand@de.ibm.com>
> Cc: gdb-patches@sourceware.org
>
> > > +@item vFile:readlink: @var{pathname}
> > > +Read value of symbolic link @var{pathname} on the target. Return
> > > +the number of bytes read, or -1 if an error occurs.
> >
> > This part is okay, but please don't use "pathname" when you really
> > mean "file name". GNU Coding Standards frown on using "path" or its
> > derivatives for anything but PATH-style directory lists.
>
> I'll be happy to use "filename" instead, but the currently existing
> packets (open, unlink) also use "pathname" today. Should those be
> changed to "filename" too?
In general, yes. But I cannot in good faith ask you to do that as
part of this patch. So let's make a first small step in this
1000-mile journey by using "filename" in just this part. I'll add to
my todo to fix the rest, if no one beats me to it.
Thanks.
> (B.t.w. note that those packets are directly related to the corresponding
> POSIX routines open/unlink/readlink -- the documentation of those routines,
> whether in POSIX itself or in the corresponding Linux man pages consistently
> refers to those arguments as "path" or "pathname" ... I'm wondering whether
> it is a deliberate decision on the part of the GNU Coding Standards to deviate
> from established terminology in that area?)
Everyone else calls the system "Linux", while the FSF insists on
"GNU/Linux". Nothing new here.