Re: next test release (5th)

["H . J . Lu" <hjl at lucon dot org>]

On Fri, Sep 29, 2000 at 05:16:05PM +0200, Jakub Jelinek wrote:
> On Fri, Sep 29, 2000 at 05:00:18PM +0200, Thorsten Kukuk wrote:
> > > This is perfectly fine.  There is no real need for using size_t.  I
> > > would perhaps just using uint32_t and int32_t instead of introducing
> > > another type.  I'll probably make this change in your patch.
> > > Otherwise people will ask why such a named type is necessry.
> > 
> > It is necessary to be binary compatible on Alpha. Or we need to
> > bump the nscd version number.
> 
> Here is an updated patch:
> 
> 2000-09-29  Jakub Jelinek  <jakub@redhat.com>
> 
> 	* nscd/nscd-client.h (NSCD_VERSION): Bump to 3.
> 	Use int32_t where appropriate.
> 	* nscd/nscd_gethst_r.c (nscd_gethst_r): Use u_int32_t where
> 	appropriate.
> 	* nscd/nscd_getgr_r.c (nscd_getgr_r): Likewise.
> 	* nscd/hstcache.c (cache_addhst): Likewise.
> 	* nscd/grpcache.c (cache_addgr): Likewise.
> 

There is a problem with this patch. When I did

# make check

on ia64 with the old nscd running, I got

socket(PF_UNIX, SOCK_STREAM, 0)         = 3
connect(3, {sin_family=AF_UNIX, path=" /var/run/.nscd_socket"}, 110) = 0
write(3, "\3\0\0\0\1\0\0\0\4\0\0\0", 12) = 12
write(3, "561\0", 4)                    = -1 EPIPE (Broken pipe)
--- SIGPIPE (Broken pipe) ---
+++ killed by SIGPIPE +++

I think the client should check the version before sending request.


H.J.