 | | | XMLParseException changes and creation of XMLLocator2 | XMLParseException changes and creation of XMLLocator2 2004-08-26 - By Michael Glavassevich
Hello Andy,
I'm pretty sure Ankit proposed this new interface because XNI's core and
parser packages were supposedly declared to be gold in the 2.3.0 release.
Of course 2.4.0 then immediately modified XMLLocator for exposing
encodings, as well as XMLResourceIdentifier for getting and setting
namespaces URIs.
As we've been finishing up the work on SAX 2.0.2 and DOM Level 3, it's
become apparent that XNI on its own doesn't provide all the things these
APIs require. Fortunately there are slots for Augmentations, so it's been
possible though cumbersome to do things like reporting skipped entities
[1] and whether an attribute was declared or not in the DTD (both SAX and
DOM need this) even though the XMLDocumentHandler and XMLAttributes
interfaces don't natively support these things. SAX's EntityResolver2 is
probably a good example of something that doesn't fit well in the current
framework. For instance, one of the parameters on the new resolveEntity
method is the name of the entity being resolved. You cannot get that
information from an XMLResourceIdentifier so I created a new interface in
the impl package [2]. This just doesn't feel like a good home for it
though. The request to add getCharacterOffset() to XMLLocator is for
providing the UTF-16 offset to DOM Level 3's DOMLocator.
While we're on the topic of changing XMLLocator, what about adding
getVersion()? Yes, I'm aware this was brought up before but now Xerces is
actually going to be implementing SAX2 Extensions 1.1. It would also be
nice if we could finalize the grammar interfaces
(org.apache.xerces.xni.grammars). I think they're pretty stable now.
Thanks.
[1]
http://xml.apache.org/xerces2-j/javadocs/api/org/xml/sax/ContentHandler.html
#skippedEntity(java.lang.String  )
[2]
http://cvs.apache.org/viewcvs.cgi/xml-xerces/java/src/org/apache/xerces/impl
/XMLEntityDescription.java?view=markup
[3]
http://www.w3.org/TR/2004/REC-DOM-Level-3-Core-20040407/core.html#Interfaces
-DOMLocator
Andy Clark <andy@(protected)> wrote on 08/26/2004 02:17:00 PM:
> Ankit Pasricha wrote:
> > The two code changes are listed below:
> >
> > XMLLocator2 extends XMLLocator {
> >
> > public int getCharacterOffset();
> > }
>
> Why are you extending the interface instead of just adding the
> method to the existing interface?
>
> I know I'd have to update some of my XNI tools to add the method
> but that XMLLocator2 just sticks out like a sore thumb. Especially
> since it's the only interface that uses this naming convention. And
> it's not the approach we've used in the past to update XNI interfaces
> -- previously, we just changed the original interface.
>
> This type of change is a lot more graceful than when we went from
> start/endPrefixMapping to passing a namespace context object to the
> startDocument method...
>
> --
> Andy Clark * andyc@(protected)
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: xerces-j-dev-unsubscribe@(protected)
> For additional commands, e-mail: xerces-j-dev-help@(protected)
>
Michael Glavassevich
XML Parser Development
IBM Toronto Lab
E-mail: mrglavas@(protected)
E-mail: mrglavas@(protected)
<br><font size=2><tt>Hello Andy,</tt></font>
<br>
<br><font size=2><tt>I'm pretty sure Ankit proposed this new interface
because XNI's core and parser packages were supposedly declared to be gold
in the 2.3.0 release. Of course 2.4.0 then immediately modified XMLLocator
for exposing encodings, as well as XMLResourceIdentifier for getting and
setting namespaces URIs.</tt></font>
<br>
<br><font size=2><tt>As we've been finishing up the work on SAX 2.0.2 and
DOM Level 3, it's become apparent that XNI on its own doesn't provide all
the things these APIs require. Fortunately there are slots for Augmentations,
so it's been possible though cumbersome to do things like reporting skipped
entities [1] and whether an attribute was declared or not in the DTD (both
SAX and DOM need this) even though the XMLDocumentHandler and XMLAttributes
interfaces don't natively support these things. SAX's EntityResolver2 is
probably a good example of something that doesn't fit well in the current
framework. For instance, one of the parameters on the new resolveEntity
method is the name of the entity being resolved. You cannot get that information
from an XMLResourceIdentifier so I created a new interface in the impl
package [2]. This just doesn't feel like a good home for it though. The
request to add getCharacterOffset() to XMLLocator is for providing the
UTF-16 offset to DOM Level 3's DOMLocator.</tt></font>
<br>
<br><font size=2><tt>While we're on the topic of changing XMLLocator, what
about adding getVersion()? Yes, I'm aware this was brought up before but
now Xerces is actually going to be implementing SAX2 Extensions 1.1. It
would also be nice if we could finalize the grammar interfaces (org.apache
.xerces.xni.grammars).
I think they're pretty stable now.</tt></font>
<br>
<br><font size=2><tt>Thanks.</tt></font>
<br>
<br><font size=2><tt>[1] http://xml.apache.org/xerces2-j/javadocs/api/org/xml
/sax/ContentHandler.html#skippedEntity(java.lang.String )</tt></font>
<br><font size=2><tt>[2] http://cvs.apache.org/viewcvs.cgi/xml-xerces/java/src
/org/apache/xerces/impl/XMLEntityDescription.java?view=markup</tt></font>
<br><font size=2><tt>[3] http://www.w3.org/TR/2004/REC-DOM-Level-3-Core
-20040407/core.html#Interfaces-DOMLocator</tt></font>
<br>
<br><font size=2><tt>Andy Clark <andy@(protected)> wrote on 08/26/2004
02:17:00 PM:<br>
<br>
> Ankit Pasricha wrote:<br>
> > The two code changes are listed below:<br>
> > <br>
> > XMLLocator2 extends XMLLocator {<br>
> > <br>
> > public int getCharacterOffset();<br>
> > }<br>
> <br>
> Why are you extending the interface instead of just adding the<br>
> method to the existing interface?<br>
> <br>
> I know I'd have to update some of my XNI tools to add the method<br>
> but that XMLLocator2 just sticks out like a sore thumb. Especially<br>
> since it's the only interface that uses this naming convention. And<br>
> it's not the approach we've used in the past to update XNI interfaces<br>
> -- previously, we just changed the original interface.<br>
> <br>
> This type of change is a lot more graceful than when we went from<br>
> start/endPrefixMapping to passing a namespace context object to the<br>
> startDocument method...<br>
> <br>
> -- <br>
> Andy Clark * andyc@(protected)<br>
> <br>
> ---------------------------------------------------------------------<br>
> To unsubscribe, e-mail: xerces-j-dev-unsubscribe@(protected)<br>
> For additional commands, e-mail: xerces-j-dev-help@(protected)<br>
> <br>
</tt></font>
<br><font size=2><tt>Michael Glavassevich<br>
XML Parser Development<br>
IBM Toronto Lab<br>
E-mail: mrglavas@(protected)</tt></font>
<br><font size=2><tt>E-mail: mrglavas@(protected)</tt></font>
|
|
|
