Den 31. mai. 2006 kl. 04.10 skrev David Crossley:
Sjur N. Moshagen (JIRA) wrote:
Sjur N. Moshagen commented on FOR-435:
--------------------------------------
Earlier investigation in our project has shown that the Chaperon
grammar is using the default Java file encoding when reading
files, and that the default Java encoding is given by the OS, in
our case MacOS X, which has MacRoman as default. Reading UTF-8
encoded files as MacRoman will of course garble non-ASCII characters.
Perhaps this issue also affects the core of Forrest.
IIUC from our sitemaps, we use Chaperon to extract
links from CSS files.
It sounds at least like a potential source of problems.
Today I put some effort into finding a work-around based on this
insight, and the result is the following command line argument:
forrest run -Dforrest.jvmargs="-Dfile.encoding=utf-8"
Perhaps this should be an available forrest property.
That would be very nice, although it should be made clear in the
documentation that it can affect more than Chaperon. The parameter
overrides the OS-provided default file encoding, and sets the
specified file encoding as default for the Java VM. Thus, all file
readers not specifying the encoding will use it.
It doesn't really solve the underlying problem of configuring
Chaperon from within Forrest (or Cocoon), but it does solve our
actual problem through a work-around.
Does the Cocoon chaperon block need some configurability
added?
AFAIR (it is a long time since I tried this), the Chaperon
documentation claims the file reading encoding to be configurable,
but I could not get it to work. Whether that was my mistake or a bug
in Chaperon is beyond me:-)
Also does our Chaperon jar need updating?
You mentioned an important mail thread below, but could
not provide the link at the time.
The link is provided in the first comment in the issue, just below
the "empty link" text.
Thanks very much for you investigation and other effort.
Thank you (and all the others) for your work with Forrest!
-David
Sjur
Wiki input files (*.jspwiki) is not correctly read when in UTF-8
----------------------------------------------------------------
Key: FOR-435
URL: http://issues.apache.org/jira/browse/FOR-435
Project: Forrest
Type: Bug
Components: Plugin: input.wiki
Versions: 0.8-dev, 0.7
Environment: MacOS X, 10.3.8, Java 1.4.2
Reporter: Sjur N. Moshagen
According to the documentation at:
http://chaperon.sourceforge.net/using-cocoon.html
it should be possible to configure the Wiki plugin (or any plugin
based on Chaperon) for different encodings of the input file, in
my case UTF-8.
But this does not work. I have:
<map:transformer name="lexer"
src="org.apache.cocoon.transformation.LexicalTransformer"
logger="sitemap.transformer.lexer">
<map:parameter name="localizable" value="true"/>
<map:parameter name="encoding" value="UTF-8"/>
</map:transformer>
in the input.xmap file in $FORREST_HOME/plugins/wiki, and I have
run "ant local-deploy", but to no avail: multibyte UTF-8
sequences come out as the Latin-1 counterpart of each byte in the
sequence.
A discussion about this bug can be found at:
[mail archive not yet updated, will add link here later]
