FWD » Core Development » Database

The following example:

def temp-table tt1 field f1 as char
                   field f2 as int
                   field f3 as int
                   field f4 as logical
                   index idx f4.

def temp-table tt2 LIKE tt1
                   field f5 as char
                   index idx2 f5. 

Creates the following dmo Tt2_1, with just the index defined in by tt2 definition and the indices of the model table are completely missed out:

@Table(name = "tt8", legacy = "tt2")
@Indices(
{
   @Index(name = "idx2", legacy = "idx2", components = 
   {
      @IndexComponent(name = "f5", legacy = "f5")
   })
})
public interface Tt2_1
extends Temporary

I think I found the problem why the indices from model table are not taken into consideration, this the culprit code in SchemaDictionary.finalizeTableDefine:

Iterator<Aast> iter = indices(table);
if (!iter.hasNext())
{
   iter = indices(modelTable);
   while (iter.hasNext())
   {
      Aast next = iter.next();
      next.graftCopyTo(table, null);
    }
}

As you can see the model indices are taken into consideration only if the main temp-table don't have any indices. My proposed solution is to take indices of the modelTable id it is not null

=== modified file 'src/com/goldencode/p2j/schema/SchemaDictionary.java'
--- old/src/com/goldencode/p2j/schema/SchemaDictionary.java    2025-08-04 17:30:00 +0000
+++ new/src/com/goldencode/p2j/schema/SchemaDictionary.java    2025-09-17 18:30:34 +0000
@@ -1872,12 +1872,12 @@
             }

             Iterator<Aast> iter = indices(table);
-            if (!iter.hasNext())
+            if (modelTable != null)
             {
-               iter = indices(modelTable);
-               while (iter.hasNext())
+               Iterator<Aast> modelIter = indices(modelTable);
+               while (modelIter.hasNext())
                {
-                  Aast next = iter.next();
+                  Aast next = modelIter.next();
                   next.graftCopyTo(table, null);
                }
             }

Have to check what happens with duplicated name indices in OE.

Constantin Asofiei wrote:

Instead, please use the SchemaDictionary.getIndex(NameNode, name) or to calculate if an index exists or not.

I saw that SchemaDictionary.getIndex(NameNode, name) will return an abbreviation match. For example if I have idx on a table, it will return a match on idx1 table even if I don't have such an index on the table.

Constantin Asofiei wrote:

10590a rev 16178 has a change from 9457c 16205:

• Fixed DEF TEMP-TABLE ... LIKE ... USE-INDEX - the index lookup must be case-insensitive.

I'll put this into conversion testing with other changes.

Right fix is in 16180.

In rev 16181, it fixes some issues with the initial approach: In 'like' temp-table definitions, indexes are copied only if the model is a temp-table, the model is a perm table and no explicit indexes in the table defined 'like', or USE-INDEX is not used."

Eduard: please check your standalone tests, conversion testing looks like it passed (there are changes, but they make sense). I'll need to do runtime testing.

Constantin Asofiei wrote:

Eduard: please check your standalone tests, conversion testing looks like it passed (there are changes, but they make sense). I'll need to do runtime testing.

My standalone tests are working with rev. 16181.

Committed on 10590a, rev. 16182. Fix for #10615-47.

Eduard, see 10590a/16183 - a copy of the index AST is required when copying from the model.