--- trunk2/NEWS.pod	2004/06/17 15:13:02	369
+++ trunk2/NEWS.pod	2004/06/20 17:52:41	375
@@ -32,12 +32,21 @@
 one value) it's not a great loss. However, you will have to re-write
 your C<import_xml> files if you use lookup files to use memory lookups.
 
+Lookups B<are not> specified by C<[v900^4]> as in v1.x. Now, you have
+to use something like C<lookup{v900^4}>. Recursive lookups are also
+possible: C<lookup{foo:lookup{v900^4}}>.
+
 Memory lookups (referred as lookups for new versions of WebPAC) are now
-available always. Since whole database is available when creating
+always available. Since whole database is available when creating
 output linking should be much easier now.
 
 However, using new lookups require to define lookups in different way.
-Additional benefit of that is conditional insert into lookup using eval.
+Additional benefit of that is conditional insert into lookup using eval
+(TODO: explain new lookup format).
+
+If you have any lookups in your tag (e.g. <isis>lookup{v900^5}</isis>)
+smart delimiters will be turned off (this behaviour is consistent with
+old in v1.x if you use C<type="lookup*"> or C<filter="mem_lookup">).
 
 =head2 Field definition in C<import_xml>
 
@@ -55,3 +64,24 @@
 
 For you, the user, that will result in changes to all templates. I'm sorry
 about this, but it had to be done.
+
+=head2 Order of substitution
+
+This was never really documented, so here it is now:
+
+=over 5
+
+=item fields
+
+substitution of fields like C<v200^a> into values
+
+=item lookups
+
+next, lookups are performed (C<lookup{...}>)
+
+=item format
+
+then formats (defined by C<format_name=> and C<format_delimiter=>) are
+applied. If formats have lookups, they will be performed again.
+
+=cut