This is the mail archive of the
libc-alpha@sourceware.org
mailing list for the glibc project.
Re: Please address NI_NUMERICSCOPE issue
- From: Rich Felker <dalias at aerifal dot cx>
- To: "Joseph S. Myers" <joseph at codesourcery dot com>
- Cc: libc-alpha at sourceware dot org
- Date: Thu, 19 Jun 2014 20:52:19 -0400
- Subject: Re: Please address NI_NUMERICSCOPE issue
- Authentication-results: sourceware.org; auth=none
- References: <20121014220643 dot GA17229 at brightrain dot aerifal dot cx> <20140604025109 dot GL507 at brightrain dot aerifal dot cx> <Pine dot LNX dot 4 dot 64 dot 1406192211300 dot 2033 at digraph dot polyomino dot org dot uk>
On Thu, Jun 19, 2014 at 10:13:11PM +0000, Joseph S. Myers wrote:
> On Tue, 3 Jun 2014, Rich Felker wrote:
>
> > On Sun, Oct 14, 2012 at 06:06:43PM -0400, Rich Felker wrote:
> > > See bug #14102: http://sourceware.org/bugzilla/show_bug.cgi?id=14102
> > >
> > > Even if nobody is willing to add the functionality immediately, can we
> > > at least agree on a value it will have when it's added?
> >
> > This issue is really trivial and I've been waiting for someone to
> > address it for almost 2 years now. Is the attached patch acceptable?
>
> You need to explain why you think ignoring the flag is a valid
> implementation (or that defining it without implementing it is more
> helpful to applications, or better accords with established glibc
> conventions, than not defining it until there is an implementation).
After looking at the code recently, I think I'm actually mistaken in
claiming that it would be a nop. IMO this makes it a more serious
omission since it's impossible to get output in the standard numeric
form. I can look into what would be involved in implementing it;
hopefully it's just a single conditional, but it might be worse if
flags don't get propagated all the way down to the code that needs to
check them.
For now can we at least agree than, when this omission is fixed in
glibc, NI_NUMERICSCOPE will have the value 256 (presently the next
unused bit)? I've already implemented it in musl with that value and
would like to keep the interface as compatible as possible.
Rich