FWD » Core Development » Database

Missing attributes:

• <table_handle>:ORIGIN-HANDLE (none/none)

Actually the support for this attribute is 90% implemented. What is missing is the 'reset'. It seems like the rules/gaps/expressions.rules is not up-to-date.
Here is a testcase:

define temp-table src1 like book.
create src1.
    src1.book-id = 10.
    src1.book-title = "Tom Sailer".
    src1.publisher = "RAFT Publishing".
    src1.isbn = "123-12-1234-123-456".
    src1.cost = 21.35.
    src1.price = 21.57.
    src1.pub-date = today - 1234.
    src1.author-id = 1010.
release src1.

define temp-table tt1 like book before-table bt1.
define dataset ds1 for tt1.
define data-source dsrc1 for src1.
buffer tt1:attach-data-source(data-source dsrc1:handle).
dataset ds1:fill.
message 1 temp-table tt1:handle
          temp-table bt1:handle
          temp-table src1:handle.

temp-table tt1:tracking-changes = yes.
buffer tt1:find-first().
tt1.book-title = "Hop-Top".
buffer tt1:buffer-release().
temp-table tt1:tracking-changes = no.

define variable chg as handle.
create dataset chg.
chg:create-like(dataset ds1:handle).
chg:get-changes(dataset ds1:handle).
message 3 chg:get-buffer-handle(1):table-handle:ORIGIN-HANDLE.

// dataset ds1:EMPTY-DATASET(). -- no change here
chg:EMPTY-DATASET().
chg:get-buffer-handle(1):EMPTY-TEMP-TABLE().
message 4 chg:get-buffer-handle(1):table-handle:ORIGIN-HANDLE.

1 1000 1001 1002
3 1000
4 ?

1 1000 1001 1002
3 1000
4 1000

Actually, there is a comment in Dataset.emptyDataset() which documents the deviation.

The following patch fixes the issue:

--- old/src/com/goldencode/p2j/persist/AbstractTempTable.java
+++ new/src/com/goldencode/p2j/persist/AbstractTempTable.java
@@ -987,6 +987,7 @@
       }

       deleteAllImpl();
+      originTable = null; 

       return logical.of(true);
    }

--- old/p2j/rules/gaps/expressions.rules
+++ new/p2j/rules/gaps/expressions.rules
@@ -247,6 +247,7 @@
 ** 067 HC  20240630 Implemented support for WEB-FILE-UPLOAD statement.
 ** 068 EM  20241220 Implemented support for SESSION:STREAM.
 ** 069 DMM 20250129 Implemented support for RCODE-INFO:MD5-VALUE.
+** 070 OM  20250204 Updated gap info for ORIGIN-HANDLE.
 */
  -->

@@ -1388,7 +1389,7 @@
          <rule>attrs.put(prog.kw_ole_naml, rw.cvt_lvl_none       | rw.rt_lvl_none)</rule> <!-- undocumented, found in possenet appbuilder source code -->
          <rule>attrs.put(prog.kw_on_fr_b , rw.cvt_lvl_none       | rw.rt_lvl_stub)</rule> <!-- on-frame-border -->
          <rule>attrs.put(prog.kw_ordinal,  rw.cvt_lvl_full       | rw.rt_lvl_full)</rule> <!-- ordinal -->
-         <rule>attrs.put(prog.kw_org_hand, rw.cvt_lvl_none       | rw.rt_lvl_none)</rule> <!-- origin-handle -->
+         <rule>attrs.put(prog.kw_org_hand, rw.cvt_lvl_full       | rw.rt_lvl_full)</rule> <!-- origin-handle -->
          <rule>attrs.put(prog.kw_org_roid, rw.cvt_lvl_full       | rw.rt_lvl_full)</rule> <!-- origin-rowid -->
          <rule>attrs.put(prog.kw_overlay , rw.cvt_lvl_full       | rw.rt_lvl_full)</rule>
          <rule>attrs.put(prog.kw_own_doc , rw.cvt_lvl_full       | rw.rt_lvl_full)</rule>

One more thing:

In Ref manual, the attribute is reported as:

Access: Read-only

However, in my test, I added the following lines at the end of the test:

chg:get-buffer-handle(1):table-handle:ORIGIN-HANDLE = 
    chg:get-buffer-handle(1):table-handle.

message 5 chg:get-buffer-handle(1):table-handle:ORIGIN-HANDLE.

5 1003

Ovidiu, the gap markings for prog.kw_first_bu and prog.kw_first_qr both have the comment "not heavily tested". Do you know if this is still the case? I'm trying to determine whether we can consider them sufficiently tested and close this task.

The implementations for these two APIs have 2 years now and they are unchanged ever since. I think we should consider them as stable ATM, even if firstBuffer() has a TODO note that specifies that BUFFER resources might be reported in a different order than assumed at the moment of implementation. That is an usual assumption for 99.9% of methods ported from 4GL, nothing out of the ordinary.