There is yet another possibility.
project.xml can be always in ISO-8859-1 then national characters can
be used in form &#xhhh; (hex) or &#nnn; (dec).
In this case internal encoding in MavenUtils.getProjectString must be set to
UTF-16.


What is preferred solution?
Please comment.

[EMAIL PROTECTED] wrote:
The following issue has been updated:

    Updater: Norbert Pabis (mailto:[EMAIL PROTECTED])
       Date: Thu, 25 Sep 2003 9:25 AM
    Comment:
This patch resolves problems with charactes in POM that are not from ISO-8859-1.

There are two issues:
1. In MavenUtils getProjectString was using always ISO-8859-1.
That could not work with characters outside this charset.
To have that fixed I needed original project.xml encoding. Unfortunately SAX which 
hides behind bewixt and digester never share
this information. So there are several possible workarounds:
- to have a veriable <pomEncoding>
- to have a property pom.encoding
- to read encoding from several first bytes of projex.xml "by hand"
- to decide that project.xml is always in UTF-8
I chose the last option.

2. In xdoc plugin tag <parse> from jelly-tags-xml is used. This tag uses dom4j which 
has a bug in SAXReader.parse(File).
Maven depends on dom4j 1.2.8, last version is 1.4 and this bug is already fixed in CVS 
but no newer version is available.
Now I could do two things:
- change Maven dependency to dom4j-snapshot (risky)
- use <xml:parse xml="URL"> instead of <xml:parse xml="File">
I chose the last option.

In addtionn I included tests that ensure that with changing dependencies encoding issues will not be broken.

This patch will probably fix http://jira.codehaus.org/secure/ViewIssue.jspa?key=MAVEN-847 too.
Changes:
Attachment changed to encoding_problems_patch.gz
---------------------------------------------------------------------
For a full history of the issue, see:


http://jira.codehaus.org/secure/ViewIssue.jspa?key=MAVEN-382&page=history

-- Norbert Pabi?

Nobody expects the Debian Inquisition!
Our two weapons are fear and surprise... and ruthless efficiency!


--------------------------------------------------------------------- To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]



Reply via email to