Re: Local locks patch, did that go anywhere?

[Julian King <jpk28 cus cam ac uk>]

> 
> Hi Havoc,
> 
> On Mon, 2003-03-24 at 17:40, Havoc Pennington wrote:
> > Cool - here is the patch I have.
> 
> 	I believe the only part of this you need to change is:
> 
> @@ -2755,7 +2781,11 @@ gconf_orb_release (void)
>  char*
>  gconf_get_daemon_dir (void)
>  {
> -  return g_strconcat (g_get_home_dir (), "/.gconfd", NULL);
> +  if (gconf_use_local_locks ())
> +    return g_strdup_printf ("/tmp/gconfd-%s",
> +                            g_get_user_name ());
> +  else
> +    return g_strconcat (g_get_home_dir (), "/.gconfd", NULL);

Umm, could you not hardwire paths in like that please?  We did
that on a local copy, and didn't feed it back to the list because
that is such an irritating thing for everyone else when you do it.

Also, depending on the logic it would be nice if it only fell back
to this after it found out that the filesystem it had been planning
on using wasn't up to some bit of functionality (hard links, locking,
whatever) and so degraded to /tmp gracefully(ish) if needed.

>  }
>  
>  char*
> 
> 	This needs to be:
> 
> if (gconf_use_local_locks)
> 	return linc_get_tmpdir ();
> 
> 	Of course - the ORB sets that, in ORBit2/src/orb/GIOP/giop.c
> (giop_tmpdir_init) - but assuming you have initialized the ORB you
> should get a valid path back to directory the ORB has to ensure is
> secure etc.
> 
> 	HTH, it'd be great to get this in,
> 
> 	Regards,
> 
> 		Michael.
> 
> -- 
>  mmeeks gnu org  <><, Pseudo Engineer, itinerant idiot

Julian