Hi Mark,
Thanks -- I didn't have any problems applyting the patch. I had to make a
few minor changes to get the code to compile on VC6, which does not support
initialization in the class definition, so you might want to take a quick
look at the result, once I check it in.
One problem: we now seem to be failing a conformance tests, related to
relative URI resolution. I think your code is correct, but before I start
fighting over the test, I'd like to make sure I should. It has to do with
the following stylesheet:
<?xml version="1.0"?>
<xsl:stylesheet xmlns:xsl="http://www.w3.org/1999/XSL/Transform";
version="1.0">
<!-- FileName: impincl27 -->
<!-- Document: http://www.w3.org/TR/xslt -->
<!-- DocVersion: 19991116 -->
<!-- Section: 2.6.2 -->
<!-- Creator: Morris Kwan -->
<!-- Purpose: href is a URI containing the "file:" scheme part. -->
<xsl:import href="file:fragments/imp27b.xsl"/>
<xsl:output method="xml" indent="no" encoding="UTF-8"/>
<xsl:template match="/">
<out>
<xsl:apply-templates select="doc" />
</out>
</xsl:template>
</xsl:stylesheet>
Notice the href of xsl:import has a scheme. From reading RFC2396, it seems
to me this is an absolute URI:
"Relative URI references are distinguished from absolute URI in that
they do not begin with a scheme name. Instead, the scheme is inherited
from the base URI, as described in Section 5.2"
Unfortunately, it looks like the Java processors (which probably rely on
the Java URL class) think this is a relative reference. The Microsoft
processor chokes if the scheme is present, which I think is correct. I
don't mind having us fail something like this if failing is correct, but
I'd rather make the argument that it ought to be removed from the
conformance suite.
What do you think?
Thanks!
Dave
|---------+--------------------------->
| | "Mark Weaver" |
| | <[EMAIL PROTECTED]|
| | > |
| | |
| | 03/20/2003 06:52|
| | AM |
|---------+--------------------------->
>--------------------------------------------------------------------------------------------------------------------------------|
|
|
| To: "Xalan-C-Users" <[email protected]>
|
| cc: (bcc: David N Bertoni/Cambridge/IBM)
|
| Subject: RE: Upcoming Xalan-C++ 1.5 release
|
>--------------------------------------------------------------------------------------------------------------------------------|
> Mark, can you make sure your patch is compatible with the latest CVS and
> provide a CVS diff? Also, actual copies of the new source files
> instead of
> diffs (XalanParsedURI.cpp and XalanParsedURI.hpp) would be much easier to
> deal with.
>
OK, the included zip is a cvs diff against the current cvs plus the
individual files. Basically the difference was the addition of namespaces
caused the patch to fail.
If you unzip in the xml-xalan directory, then the patch should apply OK and
the new files should go in the right place. If not, I've stuck
URISupport.cpp in there as well for luck. I've also added a patch against
Projects, with the caveat that I don't have VC6 so the editing was done in
notepad.
I'm not sure how the configure system works so I don't know if other
platforms will pick up the new CPP file -- is there anything I should do to
add this?
I've also included the test app, but this has only a VC7 project file, it
did run fine after recompiling things and adding a using namespace xalan.
Mark
