[
https://issues.apache.org/jira/browse/OPENJPA-1540?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12839873#action_12839873
]
Jeremy Bauer commented on OPENJPA-1540:
---------------------------------------
Laird, I committed a change under rev 917727 (to trunk) that should fix many of
the backward compatibility issues you are seeing -- provided that you are/were
using the Informix JDBC driver. The IBM JCC driver *should* have worked without
this change. Please let me know if it helps. There will still be a problem in
the reverse mapping tool if delimited ID's are used (but I doubt this is the
case since you are running with 1.2.2 as well). In addition there may still be
a problem with index identifiers that start with a number (that does not follow
SQL naming conventions and needs special handling) . I'll work on tuning the
naming code to allow this behavior on Informix.
> XMLSchemaSerializer outputs funny values for column name attributes
> (regression from 1.2.2)
> -------------------------------------------------------------------------------------------
>
> Key: OPENJPA-1540
> URL: https://issues.apache.org/jira/browse/OPENJPA-1540
> Project: OpenJPA
> Issue Type: Bug
> Components: jdbc
> Affects Versions: 2.0.0-beta
> Reporter: Laird Nelson
> Assignee: Jeremy Bauer
>
> During a reverse mapping run, it's often common to take the output of a
> SchemaGenerator and feed it to an XMLSchemaSerializer via its addAll() method.
> When the DBDictionary that was in effect at schema generation time is one
> that supports delimiters (see InformixDictionary), then the identifiers on
> all the artifacts in the SchemaGroup housed by the SchemaGenerator have
> quotes in them.
> Consequently, when the XMLSchemaSerializer writes its schemas to disk, you
> can get strange results. Here's an example <column> fragment:
> <column name=""foobar"">
> Technically I suppose this may not be a bug in XMLSchemaSerializer, since it
> is simply serializing what it is handed. But given that other tools consume
> its XML schema files, and given that the value of name attribute will be read
> in as a String, and given that usually what happens is that a String is
> turned into a DBIdentifier and delimited/quoted as necessary, I am not sure
> that delimiters should be stored by the XMLSchemaSerializer.
> A downstream effect of this bug is that any reverse mapping that takes place
> from here involving annotations will not compile. That is, if you ask the
> ReverseMappingTool for example to produce annotated Java objects, the
> annotations will look like this:
> @Column(name=""foobar"")
> ...which will not compile.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
