On Tue, Jan 24, 2006 at 07:57:25PM +0100, Vincent Hanquez wrote:
> On Tue, Jan 24, 2006 at 08:14:40PM +0200, Muli Ben-Yehuda wrote:
> > > + /* try to use a static buffer */
> >
> > sbuf isn't static here...
>
> static as contrary to dynamicly allocated, not as a static C variable
>
> do you prefer s/static/stack/ ?
Doesn't really matter.
> > vsnprintf can return a negative error value on output failure, in
> > which case calling write is not a good idea.
>
> Did you really saw a *vsnprintf* fail ever with a negative value ?
Not that I can recall.
> I cannot find any reason that it will have an output error, which is the
> only reason listed in the manual, since the output is a string in memory.
>
> I think the return value of -1 is only related to:
> fprintf, vfprintf, printf (redirection of output), vprintf
>
> If that's not true, I think that there's lots of other place that can
> have this problem.
It's not exactly the bug of the decade, but I don't see any downsides
to handling all possible errors here since we're already checking the
return value...
> > trace("%s", "\n***\n")? it's slightly slower but more future proof.
>
> well it could not be that slower anyway actually:
>
> the algo can just strlen+strcpy the first argument for the output
> instead of going through the format string char by char for an escape.
> for a short string that doesn't really matters though.
>
> I'm happy either way, but patch welcome, although that's just nitpicking
> now ;)
Guilty as charged :-)
Cheers,
Muli
--
Muli Ben-Yehuda
http://www.mulix.org | http://mulix.livejournal.com/
_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxxxxxxxx
http://lists.xensource.com/xen-devel
|
Home