FWD » Core Development » Database

The current implementation of GET-CODEPAGE() (implemented in #3753) works with LONGCHAR but not with CLOB fields. It is supposed to work with both. As implemented, it is hard coded to call longchar.getCodePage() which is not OK. In this task, the work is to:

• make a generic method in I18NOps and accept both LONGCHAR (get-codepage(lc-var)) and CLOB (get-codepage(buffer-handle:buffer-value))
• redirect to the existing longchar.getCodePage() if the parameter is LONGCHAR
• add support for CLOB and call it if the parameter is a CLOB field

I did managed to get the testcases project setup done. Right now, TestGetCodepage.cls fails because the code page is hard-coded and it does not match with the default SESSION:CPINTERNAL (1252) code page. I will add some new checks on that test, on the next working day.

Commited revision 15910 on branch 9440a, with the changes according to the latest code review.

• this must nest the @rule@s (like if/then/(if/then/else)/else); currently is is (if/then)(if/then/else).
  

              <rule>cls == "longchar" 
                 <action>methodText = "longchar.getCodePage"</action>
              </rule>
              <rule>cls == "clob" 
                 <action>methodText = "clob.getCodePage"</action>
                 <action on="false">methodText = "I18nOps.getCodePage"</action>
              </rule>
  

  
  correct:
  

              <rule>cls == "longchar" 
                 <action>methodText = "longchar.getCodePage"</action>
                 <rule on="false">cls == "clob" 
                    <action>methodText = "clob.getCodePage"</action>
                    <action on="false">methodText = "I18nOps.getCodePage"</action>
                 </rule>
              </rule>
  

• otherwise, the code looks good; but, please use character.UNKNOWN instead of new character(); also, look into the 9060b branch, and if longchar and clob are not changed to use character.UNKNOWN or character.of or character.EMPTY_STRING instead of new character, please make these improvements in this branch.

Commited revision 15911 on branch 9440a.
I changed the builtin_functions.rules, with the changes you provided. In longchar and clob classes, there are no usages of new character(). I changed new character() call from I18nOps.getCodepage(BDT) to use character.UNKNOWN.

Commited revision 15912 on branch 9440a.
Replaced new character() with character.UNKNOWN on longchar.getCodePage().

I'm OK with the changes in rev 15912.

What standalone tests did you use for this?

I modified the /tests/base_language/builtin_functions/functionality/TestGetCodepage.cls to use SESSION:CPINTERNAL, instead of "ISO8859-1". Do I need to commit that test?

Please expand the test to also use get-codepage(buffer ttTable::fClob) (i.e. the POLY case which should convert to I18NOps.getCodePage). And a case with get-codepage(buffer ttTable::fInt) which would throw the error. Please run the test in both FWD and OE.

When using get-codepage(buffer ttTable::fInt), on OpenEdge, it does not throw any error. The returned value is ?.
Do I need to change the I18nOps.getCodePage(BDT) to not throw error when using other types of instances?

Teodor Gorghe wrote:

When using get-codepage(buffer ttTable::fInt), on OpenEdge, it does not throw any error. The returned value is ?.
Do I need to change the I18nOps.getCodePage(BDT) to not throw error when using other types of instances?

Yes, if 4GL shows that it returns something only for clob or longchar args, and unknown for any other type, then FWD must do the same.

Ok, thanks.
Related to this task, I noticed that the task status become "Internal Test".
What tests do I need to run?

Teodor Gorghe wrote:

Ok, thanks.
Related to this task, I noticed that the task status become "Internal Test".
What tests do I need to run?

Run conversion for a simple app which has get-codepage.