Jan,

are this checkin to the Feb2010 branch (changeset 1e2808a8e416) and your other checkin to the Jun2010 branch (changeset cc8b3ce4c30a) meant for these branches, only, or should they also be propagated to (merged into) the Jun2010 branch (changeset 1e2808a8e416) and default branch (both changesets).

In either case, do you take care of the respective merges (both doing and preventing the propagation require merging the heads) yourself, or do you prefer us to do them?

Thanks!

Stefan

On Tue, Jul 06, 2010 at 03:32:22PM +0200, Jan Rittinger wrote:

Changeset: 1e2808a8e416 for MonetDB URL: http://dev.monetdb.org/hg/MonetDB?cmd=changeset;node=1e2808a8e416 Modified Files: pathfinder/compiler/algebra/opt/opt_complex.c pathfinder/compiler/algebra/prop/prop_dom.c pathfinder/compiler/sql/lalg2sql.brg Branch: Feb2010 Log Message:

Various small bugfixes (in pf property inference, optimizations & SQL code generation).

diffs (63 lines):

diff -r a7a0e0845e56 -r 1e2808a8e416 pathfinder/compiler/algebra/opt/opt_complex.c --- a/pathfinder/compiler/algebra/opt/opt_complex.c Tue Jul 06 07:29:54 2010 +0200 +++ b/pathfinder/compiler/algebra/opt/opt_complex.c Tue Jul 06 15:32:06 2010 +0200 @@ -2118,20 +2118,21 @@ for (unsigned int i = 0; i < schema.count; i++) schema.items[i].name = PFord_order_col_at (p->sem.sort.sortby, i);

•             if (PFprop_ckey (p->prop, schema)) {
  

•                 *p = *PFla_rowrank (
  

•                           L(p),
  

•                           p->sem.sort.res,
  

•                           p->sem.sort.sortby);
  

•                modified = true;
  

•                break;
  

•             }
  

•            if (PFprop_ckey (p->prop, schema)) {
  

•                *p = *PFla_rowrank (
  

•                          L(p),
  

•                          p->sem.sort.res,
  

•                          p->sem.sort.sortby);
  

•               modified = true;
  

•               break;
  

•            }
         }
  
         /* Replace the rownumber operator by a projection
            if only its value distribution (keys) are required
            instead of its real values. */
         if (!PFprop_req_value_col (p->prop, p->sem.sort.res) &&
  

• !p->sem.sort.part &&
          PFord_count (p->sem.sort.sortby) == 1 &&
          PFprop_key (p->prop,
                      PFord_order_col_at (p->sem.sort.sortby, 0))) {
  

diff -r a7a0e0845e56 -r 1e2808a8e416 pathfinder/compiler/algebra/prop/prop_dom.c --- a/pathfinder/compiler/algebra/prop/prop_dom.c Tue Jul 06 07:29:54 2010 +0200 +++ b/pathfinder/compiler/algebra/prop/prop_dom.c Tue Jul 06 15:32:06 2010 +0200 @@ -907,9 +907,11 @@ L(n)->schema.items[i].name), dom);

•            add_disjdom (dom,
  

•                         PFprop_dom (R(n)->prop,
  

•                                     L(n)->schema.items[i].name));
  

•            /* we can only infer the disjointness for a single column */
  

•            if (n->schema.count == 1)
  

•                add_disjdom (dom,
  

•                             PFprop_dom (R(n)->prop,
  

•                                         L(n)->schema.items[i].name));
         }
         break;
  
  

diff -r a7a0e0845e56 -r 1e2808a8e416 pathfinder/compiler/sql/lalg2sql.brg --- a/pathfinder/compiler/sql/lalg2sql.brg Tue Jul 06 07:29:54 2010 +0200 +++ b/pathfinder/compiler/sql/lalg2sql.brg Tue Jul 06 15:32:06 2010 +0200 @@ -4078,7 +4078,7 @@ } /* In case the column did not exist in the column list already ... */

•                    if (!found)
  

•                    if (!found && !IS_LITERAL(expr))
                         /* ... extend the group by list. */
                         groupbylist = column_list (groupbylist, expr);
                 }   break;
  

---

Checkin-list mailing list Checkin-list@monetdb.org http://mail.monetdb.org/mailman/listinfo/checkin-list