[
https://issues.apache.org/jira/browse/XALANJ-2618?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17811343#comment-17811343
]
Joe Kesselman commented on XALANJ-2618:
---------------------------------------
The two hashtables suggest that we originally thought we might be mapping from
Java name (or java-ish name) to mime names as well as vice versa. The question
is whether anything is relying on that, and if so whether it is also relying on
our legacy support for the shortened names. I wouldn't want to remove something
that might cause a behavior change visible to the user, if at all avoidable.
One workaround might be to drop the separate listings of ISOxxxx-x versus
xxxx-x, and have code going that direction search twice, once as provided and
(if that fails) once with the ISO prefix inserted.. That would retain the
ability to use the short name when going from java to mime, while removing it
from the MIME-to-Java map. Lookup would be a bit more expensive, admittedly.
> Error in org/apache/xml/serializer/Encodings.properties
> -------------------------------------------------------
>
> Key: XALANJ-2618
> URL: https://issues.apache.org/jira/browse/XALANJ-2618
> Project: XalanJ2
> Issue Type: Bug
> Security Level: No security risk; visible to anyone(Ordinary problems in
> Xalan projects. Anybody can view the issue.)
> Components: Serialization, transformation
> Affects Versions: 2.7.2
> Environment: Java 11
> Reporter: Simon Schaarschmidt
> Assignee: Steven J. Hathaway
> Priority: Major
> Labels: Java11
>
> We transform and serialize using encoding ISO-8859-1. With JDK 1.8 all is
> fine, but with OpenJDK 11 the result will be written (from class
> ToTextStream) in character references, e.g.
> "*id=1*" instead of "*id=1*".
> In org/apache/xml/serializer/Encodings.properties (serializer.jar) are
> various encodings defined, e.g.
> {{ISO8859-1 ISO-8859-1 0x00FF}}
> {{ISO8859_1 ISO-8859-1 0x00FF}}
> {{{color:#ff0000}8859-1{color} ISO-8859-1 0x00FF}}
> {{{color:#ff0000}8859_1{color} ISO-8859-1 0x00FF}}
> First value: Java encoding name
> Second value: comma separated preferred mime names.
> The class org.apache.xml.serializer.Encodings reads this file in a Properties
> object and processes the definitions to create EncodingInfo objects and puts
> them (see method loadEncodingInfo()) into the member fields
> __encodingTableKeyJava_ and __encodingTableKeyMime_ (both Hashtable).
> Especially putting Elements into _encodingTableKeyMime is critical because
> there is not a 1:1 mapping and the latest returned Properties.keys() element
> replaces the previous ElementInfo object.
> Until Java 1.8 the first line from above is the latest entry in Enumeration,
> therefor _encodingTableKeyMime returns the EncodingInfo object with Java
> encoding "{color:#14892c}ISO8859-1{color}" for encoding "ISO-8859-1". With
> Java 11 the elements of the Enumeration returned by Properties.keys() has a
> different order: the third line from above is the latest entry! Therefor
> _encodingTableKeyMime returns the EncodingInfo object with Java encoding
> "*{color:#ff0000}8859-1{color}*" when asking for encoding "ISO-8859-1". But:
> "8859-1" ist not a valid Java encoding name! Method
> EncodingInfo.inEncoding(char,String) fails internally with an
> *UnsupportedEncodingException* and returns false.
> The methods in class Encodings first searches EncodingInfo object in
> _encodingTableKeyJava and uses elements from _encodingTableKeyMime as
> fallback.
> I suggest the definitions in Encodings.properties must be extended with
> additional lines, e.g.
> {{*{color:#14892c}ISO-8859-1{color}* ISO-8859-1 0x00FF}}
> Also for encodings ISO-8859-2..9. Or all entries with Java encoding name
> "8859*" should be removed. (They are not valid Java encoding names -
> UnsupportedEncodingException!)
> Finally I think, the current mechanism of collecting the EncodingInfo objects
> using two Hashtables is critical.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscr...@xalan.apache.org
For additional commands, e-mail: dev-h...@xalan.apache.org
