You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@ant.apache.org by Antoine Levy-Lambert <an...@gmx.de> on 2005/12/07 09:36:19 UTC
URI encoding ...
I have checked in some code lately related to URI encoding and decoding.
I was triggered by 2 bug reports, which were pointing to the same
problems in the background.
One bug report was about the impossibility to load properly a jar file
having in his manifest a Class-Path attribute pointing to another jar
file in a subdir, when the main jar file is in a directory containing
spaces.
The other bug report was about entities not found when processing with
<xslt/> an XML document located in a non ascii path and referring to
entities in the same directory.
Now I am testing manifestclasspath. I tested it with änt (a umlaut nt,
or ã) . this works. Now with a directory bearing a hebrew name (iom
= םוי ) it does not work.
I can send a tar file with my test material if someone is interested. I
am wondering whether our encoding/decoding routines are wrong, or
whether it is a specific problem for langages which get written right to
left or
could even be some kind of JDK bug ...
Cheers,
Antoine
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@ant.apache.org
For additional commands, e-mail: dev-help@ant.apache.org
Re: URI encoding ...
Posted by Antoine Levy-Lambert <an...@gmx.de>.
Jesse Glick wrote:
> Antoine Levy-Lambert wrote:
>
>> I have checked in some code lately related to URI encoding and decoding.
>
>
> BTW should use java.net.URI and File.toURI and new File(URI) when
> running on JDK 1.4+. I have an old outstanding bug for this, needs to
> be reexamined:
>
> http://issues.apache.org/bugzilla/show_bug.cgi?id=8031
Will look at this.By the way, the methods to encode/decode URIs are
available in JDK 1.5, for instance URLEncoder.encode("UTF-8") does the
same as org.apache.tools.ant.launch.Locator#encodeUri,
>
>> One bug report was about the impossibility to load properly a jar file
>> having in his manifest a Class-Path attribute pointing to another jar
>> file in a subdir, when the main jar file is in a directory containing
>> spaces.
>
>
> Note that a relative path in Class-Path is really a URI and must be
> encoded. E.g.
>
> Class-Path: some%20other%20lib.jar
Yes; the % escaping of the class path is now done in the new
manifestclasspath task.
AntClassloader
>
>> The other bug report was about entities not found when processing with
>> <xslt/> an XML document located in a non ascii path and referring to
>> entities in the same directory.
>
>
> May be completely unrelated, but I recently found a JDK bug sounding
> very similar to this:
>
> http://bugs.sun.com/bugdatabase/view_bug.do?bug_id=6341770
This bug report is fixed in the head revision of
javax.xml.parsers.SAXParser
http://cvs.apache.org/viewcvs.cgi/xml-commons/java/external/src/javax/xml/parsers/SAXParser.java?rev=1.7&view=log
We just have to light a candle, hoping that the new version of SAXParser
will be in the new JDK 1.5s shipped by Sun.
>
>> Now I am testing manifestclasspath. I tested it with änt (a umlaut nt,
>> or ã) . this works. Now with a directory bearing a hebrew name (iom
>> = םוי ) it does not work.
>
>
> Just a guess: maybe you forgot to encode in UTF-8 before encoding to
> %xx octets? That could cause problems for non-ISO-8859-1 text.
No, the routines in Locator now do this UTF-8 encoding. I do not know
yet what is wrong. I am going to test this on Solaris to see whether
this problem could be a Windows-JVM bug.
>
> -J.
>
Cheers,
Antoine
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@ant.apache.org
For additional commands, e-mail: dev-help@ant.apache.org
Re: URI encoding ...
Posted by Jesse Glick <je...@sun.com>.
Antoine Levy-Lambert wrote:
> I have checked in some code lately related to URI encoding and decoding.
BTW should use java.net.URI and File.toURI and new File(URI) when
running on JDK 1.4+. I have an old outstanding bug for this, needs to be
reexamined:
http://issues.apache.org/bugzilla/show_bug.cgi?id=8031
> One bug report was about the impossibility to load properly a jar file
> having in his manifest a Class-Path attribute pointing to another jar
> file in a subdir, when the main jar file is in a directory containing
> spaces.
Note that a relative path in Class-Path is really a URI and must be
encoded. E.g.
Class-Path: some%20other%20lib.jar
> The other bug report was about entities not found when processing with
> <xslt/> an XML document located in a non ascii path and referring to
> entities in the same directory.
May be completely unrelated, but I recently found a JDK bug sounding
very similar to this:
http://bugs.sun.com/bugdatabase/view_bug.do?bug_id=6341770
> Now I am testing manifestclasspath. I tested it with änt (a umlaut nt,
> or ã) . this works. Now with a directory bearing a hebrew name (iom
> = םוי ) it does not work.
Just a guess: maybe you forgot to encode in UTF-8 before encoding to %xx
octets? That could cause problems for non-ISO-8859-1 text.
-J.
--
jesse.glick@sun.com x22801 netbeans.org ant.apache.org
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@ant.apache.org
For additional commands, e-mail: dev-help@ant.apache.org