[PATCH] Re: resolv_wrapper URI RR support - name compression

[Nathaniel McCallum]

On Wed, 2016-08-24 at 10:53 -0400, Matt Rogers wrote:
> On Wed, 2016-08-24 at 08:10 +0200, Andreas Schneider wrote:
> 
> > On Tuesday, 23 August 2016 11:41:45 CEST Matt Rogers wrote:
> > 
> > > 
> > > On Tue, 2016-08-23 at 15:54 +0200, Andreas Schneider wrote:
> > > 
> > > > 
> > > > On Monday, 22 August 2016 17:52:25 CEST Matt Rogers wrote:
> > > > 
> > > > > 
> > > > > Hi,
> > > > > 
> > > > > I'm using resolv_wrapper from master at 02535da4 to test krb5
> > > > > URI
> > > > > RR
> > > > > lookups. Similar to the SRV records, rwrap_fake_uri() runs
> > > > > ns_name_compress() on the result, however URI records contain
> > > > > a
> > > > > formatted entry and not a DNS name, so I don't think that the
> > > > > call
> > > > > to
> > > > > ns_name_compress() is valid here.  The result of running
> > > > > ns_name_compress() on the URI record (which appears as
> > > > > "krb5srv:m:kkdcp:https://kdc/path") is that it adds a " " to
> > > > > the
> > > > > beginning that does not get removed with a subsequent
> > > > > uncompress.
> > > > > Any
> > > > > objections to removing the ns_name_compress/uncompress on URI
> > > > > RR
> > > > > results?
> > > > 
> > > > This means the provided test is incorrect. We need a test which
> > > > works
> > > > in a 
> > > > real world scenario and also with resolv_wrapper to verify that
> > > > resolv_wrapper 
> > > > is doing everything correctly.
> > > > 
> > > > Could you use the current test, modify it and make sure it
> > > > works
> > > > in
> > > > the real 
> > > > world?
> > > > 
> > > > Then we can fix resolve_wrapper ...
> > > 
> > > I've attached patches for removing the name compression from the
> > > URI
> > > result as well as the tests. This gives a correct result for the
> > > real
> > > world test.
> > 
> > Hey Matt,
> > 
> > thanks for the patches!
> > 
> > Please add:
> > 
> > size_t uri_len;
> > 
> > uri_len = strlen(rr->rrdata.uri_rec.uri) + 1;
> > 
> > and then use uri_len as we are using it twice.
> > 
> > Also you need to sign your work. See
> > 
> > https://www.samba.org/samba/devel/copyright-policy.html
> > 
> 
> Thanks, I've attached the new patches.

LGTM. Thanks for fixing my bug Matt!