FWD » Core Development » Database

Failing test case is dbuf-query-extent1-dt.p (Stanislav: please provide test case; it is not currently in the testcases project).

Hibernate error is:

[01/23/2014 19:54:50 TMT] (com.goldencode.p2j.persist.Persistence:WARNING) [00000001:00000007:syman-->local/_temp/primary] error executing query [select tt__1.id from DynamicRecord1Impl as tt__1 where ((tt__1._multiplex = ?) and (tt__1.field1[2] = 3)) order by tt__1._multiplex asc, tt__1.id asc]

My initial guess of the cause is that we are bypassing the HQLPreprocessor, which would normally rewrite the HQL. See the discussion of this issue in #1652, starting at note 118.

I'll post the testcase here.

def var th as handle.
def var bh as handle.
def var qh as handle.

create temp-table th.
th:undo = true.
th:add-new-field("f1", "integer", 3).
th:temp-table-prepare("tt").

bh = th:default-buffer-handle.
bh:buffer-create.
bh:buffer-field("f1"):buffer-value(1) = 1.
bh:buffer-field("f1"):buffer-value(2) = 2.
bh:buffer-field("f1"):buffer-value(3) = 3.
bh:buffer-release.

create query qh.
qh:set-buffers(bh).
qh:query-prepare("for each tt where tt.f1[3] = 3").
qh:query-open.
qh:get-next.
message string(bh:buffer-field("f1"):buffer-value(2)).

The issue was related to the fact that entries for dynamic tables are missing in DatabaseManager.composites. DatabaseManager.dynamicComposites context-local map was added to address this.

Code review 20140205a:

In general, looks good. Please regression test.

There are some new TODOs added to TemporaryBuffer.copyAllRows. What is the plan for those?

I noticed one case of new HashMap() (i.e., no "<>") (line 1759 in TemporaryBuffer). Does that produce a compiler warning? If so, please clean it up.

There are some new TODOs added to TemporaryBuffer.copyAllRows. What is the plan for those?

1.

// TODO in copy-temp-table sorting order affects which record remains in constraints
// violation case. The correct sorting is by primary index if it presents or by
// the index with the lexicographically first name.

As far as I've tested it affects only COPY-TEMP-TABLE, so no plans.
2.

// TODO also validate against already copied records

You should decide if the case when we have two tables with the same structure but different unique indexes during TABLE PARAMETER processing should be addressed now (I have a similar note in TABLE PARAMETER thread).