[
https://issues.apache.org/jira/browse/DAFFODIL-1204?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17264127#comment-17264127
]
Steve Lawrence commented on DAFFODIL-1204:
------------------------------------------
So that error is from Daffodil, whih probably does mean that the Scala XML
library did actually fix this.
But it looks like Daffodil now dosn't allow these special characters. Looking
at the QNameRegex object, we build a regex to match QNames. But there are a
handful of match patterns commented out, and TODO comments like "Why is this
one left out?", including a xB7 which is what this tests uses.
We need to scrutinize the QNameRegex and determine if certain parts should be
uncommented based on what should be allowed in the XSD specification.
For reference: https://www.w3.org/TR/xml-names11/#ns-decl
> Unicode U+00B7 character not allowed in namespace prefix
> --------------------------------------------------------
>
> Key: DAFFODIL-1204
> URL: https://issues.apache.org/jira/browse/DAFFODIL-1204
> Project: Daffodil
> Issue Type: Bug
> Components: Front End
> Reporter: Mike Beckerle
> Assignee: IanCarlson
> Priority: Minor
> Attachments: test_namespaceSpecialChars_trace.txt
>
>
> Some changes in 0.16.0 have us using the scala ConstructingParser to parse
> DFDL schemas now.
> A small incompatibility with the prior use of the scala basic XML loader was
> found.
> Test test_namespaceSpecialChars() moved to scala debug.
> The problem is that the Uncode codepoint U+00B7 is not properly accepted as
> part of a namespace prefix, so the test fails now.
> This was reported as a bug in the scala library.
> https://issues.scala-lang.org/browse/SI-9060
> The issue that has us switching from basic XML loader to ConstructingParser
> is unrelated to this small regression It is CDATA region preservation that
> has us switching parsers. This problem looks like it is getting solved in
> Scala 2.11.7, which we'll switch to some day, allowing us to go back to the
> basic scala XML loader - and that would then fix this issue with the Unicode
> B7 codepoint.
> So I suggest no action to address this bug.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
