FWD » Core Development » Base Language

Because I didn't find any test class that tests the functionality of PUT-DOUBLE statement, I've created a new class TestPutDouble.cls that contains more @Test methods.

The problem is that this tests are working in 4GL, but after I've converted the code and run it, a part of them don't work anymore.
I noticed the problem comes from the message of the errors.
For example, in 4GL, this code is working:

    @Test.
     method public void testFunctionMemptrPositionZero():
        define variable mem as memptr no-undo.

        // Position is 0
        set-size(mem) = 10.
        put-double(mem, 0) = 2.3 no-error.
        support.test.AssertExt:Error(1830, "Byte position for PUT-BYTE, GET-BYTE and other raw functions must be 1 or greater. (1830)").

     end method.

but after I converted it and run it, I receive this error:
testFunctionMemptrPositionZero ✘ Statement expected to throw error with message like 'Byte position for PUT-BYTE, GET-BYTE and other raw functions must be 1 or greater. (1830)' at index 1 but got '** Byte position for PUT-BYTE, GET-BYTE and other raw functions must be 1 or greater. (1830)'.

The converted java code for this method is:

@LegacySignature(type = Type.METHOD, name = "testFunctionMemptrPositionZero")
   @Test(order = 4)
   public void testFunctionMemptrPositionZero()
   {
      memptr mem = TypeFactory.memptr();

      internalProcedure(TestPutDouble.class, this, "testFunctionMemptrPositionZero", new Block((Body) () -> 
      {
         //  Position is 0
         mem.setLength(10);
         silent(() -> mem.setDouble(decimal.fromLiteral("2.3"), 0));
         com.goldencode.testcases.support.test.AssertExt.error(new integer(1830), new character("Byte position for PUT-BYTE, GET-BYTE and other raw functions must be 1 or greater. (1830)"));
      }));
   }

I have the same problem with error: "Unable to evaluate expression with UNKNOWN value in argument. (4391)".

Also, I run an already written test class: /home/dmm/gcd/testcases/testcases/src/com/goldencode/testcases/TestGetDouble.java and for the "Unable to evaluate expression with UNKNOWN value in argument. (4391)" error, I receive:
testFunctionRaw ✘ Statement expected to throw error with message like 'Unable to evaluate expression with UNKNOWN value in argument. (4391)' at index 1 but got '** Unable to evaluate expression with UNKNOWN value in argument. (4391)'.

Should I consider this tests as passed or should I look into it?

Committed my solution to testcases
I've added a new test class tests.base_language.builtin_functions.functionality.TestPutDouble.cls that tests PUT-DOUBLE functionality.

Committed my solution for errors at running to 9446a revision 15653.

I modified BinaryData.java file in order to solve the errors mentioned before. In genIndexError and genUnknownArgError methods, I modified the calls to ErrorManager.recordOrThrowError method by setting prefix parameter to false. This parameter, if is set to true, causes "** " to be appended to the original error text in buildErrorText method from ErrorManager.java class.

Initial code:

public static void genIndexError()
   throws ErrorConditionException
   {
      String err = "Byte position for PUT-BYTE, GET-BYTE and other" +
                   " raw functions must be 1 or greater";

      ErrorManager.recordOrThrowError(1830, err);
   }

private void genUnknownArgError()
   throws ErrorConditionException
   {
      String err = "Unable to evaluate expression with UNKNOWN value in argument";
      ErrorManager.recordOrThrowError(4391, err);
   }

Modified code:

public static void genIndexError()
   throws ErrorConditionException
   {
      String err = "Byte position for PUT-BYTE, GET-BYTE and other" +
                   " raw functions must be 1 or greater";

      ErrorManager.recordOrThrowError(1830, err, false);
   }

private void genUnknownArgError()
   throws ErrorConditionException
   {
      String err = "Unable to evaluate expression with UNKNOWN value in argument";
      ErrorManager.recordOrThrowError(4391, err, false);
   }

There are some adjacent changes that were committed. Please change them back and make a commit to get the value back. For local development, you can make a copy of the testcases project so your personal changes won't get caught up in the BZR work.

=== modified file 'deploy/client/junit.sh' (properties changed: -x to +x)
--- old/deploy/client/junit.sh    2024-05-06 18:58:14 +0000
+++ new/deploy/client/junit.sh    2025-01-20 15:00:43 +0000
@@ -24,7 +24,7 @@
       d  ) debug=$OPTARG ;;
       s  ) suspend="y" ;;
       t  ) prog="echo "$prog ; test="y" ;;
-      c  ) class="com.goldencode.testcases.tests.$OPTARG" ;;
+      c  ) class="com.goldencode.testcases.$OPTARG" ;;
       \? ) for i in "${usage[@]}"; do
               echo -e $i
            done
@@ -60,6 +60,6 @@
 fi

 # run the jUnit
-cmd="$prog $dtxt $lpath -jar /home/rfb/projects/junit-platform-console-standalone-1.9.0.jar $cpath --select-class $class" 
+cmd="$prog $dtxt $lpath -jar /home/dmm/gcd/testcases/testcases/deploy/lib.junit/junit-platform-console-standalone-1.9.2.jar $cpath --select-class $class" 
 $cmd

=== modified file 'deploy/server/prepare_dir_postgres.json'
--- old/deploy/server/prepare_dir_postgres.json    2025-01-15 19:00:11 +0000
+++ new/deploy/server/prepare_dir_postgres.json    2025-01-20 15:00:43 +0000
@@ -27,12 +27,12 @@
   "dbnames": "tstcasesdb",
   "tstcasesdb": {
     "dbtype": "postgres",
-    "dbuser": "fwd_user",
-    "dbuserpass": "user",
+    "dbuser": "fwd_admin",
+    "dbuserpass": "admin",
     "dbadmin": "fwd_admin",
     "dbadminpass": "admin",
     "collation": "en_US@iso88591_fwd_basic",
     "dbhost": "localhost",
-    "dbport": {dbport}
+    "dbport": 5434
   }
 }

=== modified file 'prepare_build.json'
--- old/prepare_build.json    2024-12-10 20:23:03 +0000
+++ new/prepare_build.json    2025-01-20 15:00:43 +0000
@@ -2,5 +2,5 @@
    "dbh2": false,
    "dbpostgres": true,
    "dbmariadb": false,
-   "dbport": 5432
+   "dbport": 5434
 }

As for the tests: they look quite comprehensive, good job.

I would be also interested to find out how this put-double integrates with other statements, more precisely get-byte. If the put-double populates 8 bytes, then lets see how get-byte is planning to extract that double byte-by-byte. Is the representation little-endian or big-endian? How are the negative double values represented? Also, please check that putting a double "pads" the bytes till that start position with ?.

Lastly, but most important, please check whether the encoding of the double values in 4GL are IEEE-754 (. I think so, but please do some unit tests for negative numbers, positive numbers, zero, etc. to check sign (1bit), exponent (11bits), mantissa (52bits).

I am ok with the changes in 9446a.
Is there anything left to be tested?

Branch 9446a was merged into the trunk as revision 15705 and archived.