remove dead code associated with unsupported National Char implementation

• Key: DERBY-2720
• Type: Improvement
• Status: Reopened
• Priority: Minor
• Assignee: Mamta A. Satoor
• Reporter: Mike Matrigali
• Component/s: SQL
• Fix Version/s: 10.3.3.1

Description

Derby still has some untested, unused code relating to a non-standard implementation of a Nationa Char type. The current code can be removed.
I believe the interesting functionality associated with this is now provided by DERBY-1478 (territory based collation) . If Derby ever implements a
National Char type it should do so differently than the existing code, collation should not be tied to the National Char type.

I believe a future National char type might have to maintain a separate type id for compatibility with jdbc interface, but actual implmentation should be
the same code as the char types. Collating of the the national char type should be supported in exactly same way as regular char types.

If anyone is really intested in the national char code, it's history will always be available in svn, and a consistent version is available by looking at 10.0, 10.1,
and 10.2 codelines. I would propose any removal of code only take place in trunk and not be backported to a released codeline.

Comments

[Mamta A. Satoor]

I have removed the disabled national character code from trunk. The changes went in as part of revision 612525 into trunk and the tests ran fine with Sun's jdk14 on Windows XP machine. I will port these changes into 10.3 too.

[Mamta A. Satoor]

Merged changes from trunk(612525) into 10.3 codeline(revision 612576).

The checkin above caused test failure in SetObjectUnsupportedTest which runs under JDK1.6. Unfortunately, I ran my tests only with JDK1.4 and hence I didn't catch the problem before checkin. I just fixed the problem with revision 612865 in trunk. Will work on migrating it to 10.3 The commit comments for trunk were as follows

While working on DERBY-2720, I removed national datatypes as one of the unsupported data types(revision 612525) and that caused the test failure for SetObjectUnsupportedTest since we stopped recognizing national dataypes as one of the unsupported datatypes. The fix required changes in only one class which I had modified incorrectly in revision 612525. This test failure was not caught earlier by my local runs of the test because I had used JDK14 which has JDBC3 support and this paritcular test requires JDBC4 support.

[Mamta A. Satoor]

Merged the trunk revision 612865 into 10.3 codeline (revision 612872)

[Daniel John Debrunner]

Some of the code that the national types (incorrectly) used to convert between datetime values and national characters has not been removed.

E.g. the SQLChar.get{Date,Time.Timestamp}Format methods are not used (I think).
Removing these methods may show other methods not used and may progress to the date time methods in LocaleFinder being removed.

I'm testing removing the getXXXFormat() methods.