[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <20250606025613.GJ1827@brightrain.aerifal.cx>
Date: Thu, 5 Jun 2025 22:56:13 -0400
From: Rich Felker <dalias@...c.org>
To: Andy Caldwell <andy.m.caldwell@...glemail.com>
Cc: musl@...ts.openwall.com, David Steele <david@...ackrest.org>,
noloader@...il.com
Subject: Re: Sign conversion warning in FD_ISSET()
On Fri, Jun 06, 2025 at 01:34:53AM +0100, Andy Caldwell wrote:
> On Fri, Jun 6, 2025 at 1:29 AM Nick Wellnhofer <wellnhofer@...um.de> wrote:
> >
> > On Jun 6, 2025, at 02:23, Andy Caldwell <andy.m.caldwell@...glemail.com> wrote:
> > >
> > >>> There probably is some way to get it right with clever use of the
> > >>> ternary operator but I haven't come up with anything in the first few
> > >>> seconds thinking about it.
> > >>>
> > >>> Rich
> > >
> > > Yes, the ternary operator can be used to create a C89-compatible
> > > version (this allows anything coercible to an `int` or an `unsigned`,
> > > so it mostly only refuses pointers and structures).
> > >
> > > #define SOCKET_TO_UNSIGNED(s) ((unsigned)(1?(s):0))
> >
> > This doesn't work with pointers because 0 is a null pointer constant. But if you change the value to 1, it should work.
> >
> > Nick
> >
>
> Thanks Nick and Rich, you are both correct, I originally had it as
> `0?1:(s)` then changed it (without re-testing) for the email, like a
> fool. However in re-testing I notice that this only triggers a
> warning `-Wint-conversion` rather than an error when passed a pointer
> in C89 mode (it gives an error when passed a structure in any mode).
Something like ((unsigned)(1?1*(s):0)) ?
Rich
Powered by blists - more mailing lists
Confused about mailing lists and their use?
Read about mailing lists on Wikipedia
and check out these
guidelines on proper formatting of your messages.
