OrientDB having trouble with Unicode, Turkish, and enums

Question

I am using a lib which has an enum type with consts like these;

Type.SHORT
Type.LONG
Type.FLOAT
Type.STRING

While I am debugging in Eclipse, I got an error:

No enum const class Type.STRİNG

As I am using a Turkish system, there is a problem on working i>İ but as this is an enum const, even though I put every attributes as UTF-8, nothing could get that STRING is what Eclipse should look for. But it still looks for STRİNG and it can't find and I can't use that. What must I do for that?

Project > Properties > Resouce > Text file encoding is UTF-8 now. Problem keeps.

EDIT: More information may give some clues which I can't get; I am working on OrientDB. This is my first attempt, so I don't know if the problem could be on OrientDB packages. But I am using many other libs, I have never seen such a problem. There is a OType enum in this package, and I am only trying to connect to the database.

    String url = "local:database";
    ODatabaseObjectTx db = new ODatabaseObjectTx(url).
    Person person = new Person("John");
    db.save(person);
    db.close();

There is no more code I use yet. Database created but then I get the java.lang.IllegalArgumentException:

Caused by: java.lang.IllegalArgumentException: No enum const class com.orientechnologies.orient.core.metadata.schema.OType.STRİNG
    at java.lang.Enum.valueOf(Unknown Source)
    at com.orientechnologies.orient.core.metadata.schema.OType.valueOf(OType.java:41)
    at com.orientechnologies.orient.core.sql.OCommandExecutorSQLCreateProperty.parse(OCommandExecutorSQLCreateProperty.java:81)
    at com.orientechnologies.orient.core.sql.OCommandExecutorSQLCreateProperty.parse(OCommandExecutorSQLCreateProperty.java:35)
    at com.orientechnologies.orient.core.sql.OCommandExecutorSQLDelegate.parse(OCommandExecutorSQLDelegate.java:43)
    at com.orientechnologies.orient.core.sql.OCommandExecutorSQLDelegate.parse(OCommandExecutorSQLDelegate.java:28)
    at com.orientechnologies.orient.core.storage.OStorageEmbedded.command(OStorageEmbedded.java:63)
    at com.orientechnologies.orient.core.command.OCommandRequestTextAbstract.execute(OCommandRequestTextAbstract.java:63)
    at com.orientechnologies.orient.core.metadata.schema.OClassImpl.addProperty(OClassImpl.java:342)
    at com.orientechnologies.orient.core.metadata.schema.OClassImpl.createProperty(OClassImpl.java:258)
    at com.orientechnologies.orient.core.metadata.security.OSecurityShared.create(OSecurityShared.java:177)
    at com.orientechnologies.orient.core.metadata.security.OSecurityProxy.create(OSecurityProxy.java:37)
    at com.orientechnologies.orient.core.metadata.OMetadata.create(OMetadata.java:70)
    at com.orientechnologies.orient.core.db.record.ODatabaseRecordAbstract.create(ODatabaseRecordAbstract.java:142)
    ... 4 more

Here is OType class: http://code.google.com/p/orient/source/browse/trunk/core/src/main/java/com/orientechnologies/orient/core/metadata/schema/OType.java

And other class; OCommandExecutorSQLCreateProperty: http://code.google.com/p/orient/source/browse/trunk/core/src/main/java/com/orientechnologies/orient/core/sql/OCommandExecutorSQLCreateProperty.java

Line 81 says: type = OType.valueOf(word.toString());

Answer 1

Am I correct to assume you are running this program using a turkish locale? Then it seems the bug is in line 118 of OCommandExecutorSQLCreateProperty:

linkedType = OType.valueOf(linked.toUpperCase());

You would have to specify the Locale whose upper casing rules should be used, probably Locale.ENGLISH as the parameter to toUpperCase.

Answer 2

This problem is related to your database connection. Presumably, there's a string in OrientDB somewhere, and you are reading it, and then trying to use it to select a member of the enum.

I'm assuming in the code that you posted that the variable word comes from data in the database. If it comes from somewhere else, then the problem is the 'somewhere else'. If OrientDB, for some strange reason, returns 'STRİNG' as metadata to tell you the type of something, then that is indeed a defect in OrientDB.

If that string actually contains a İ, then no Eclipse setting will have any effect on the results. You will have to write code to normalize İ to I.

If you dump out the contents of 'word' as a sequence of hex values for the chars of the string, I think you'll see your İ staring right at you. You have to change what's in the DB to have a plain old I.