commons-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Christopher Lenz <>
Subject Re: [Digester] Namespacesupport, Patch
Date Mon, 02 Sep 2002 13:34:28 GMT
Stefan Heimann wrote:
> On Mon, Sep 02, 2002 at 02:58:26PM +0200, Christopher Lenz wrote:
>>Hi Stefan,
>>Stefan Heimann wrote:
>>>Hi all,
>>>as posted earlier, here is the patch that adds support for rules that
>>>can match element in different namespaces. For example, its possible
>>>the have a pattern ``/foo:a/bar:b'' that matches
>>>I have updated the matching behaviour of RulesBase and
>>>ExtendedBaseRules. The matching semantic did not change, there are
>>>only some more possibilities.
>>While I would love to see the functionality you've added in Digester, I 
>>think the change should be done with more care.
> Could you describe more exactly what you mean by ``with more care''?

Maybe that sounded too harsh, I didn't want to express that the code was 
not good or anything ;o)

My basic concern is that you're proposing backwards incompatible API 
changes just to replace the pattern-String with a QNameList (IIRC). My 
suggestion is to replace it with a more generic class that represents 
the input document context (I've called it DigesterContext in my 
example). The reasoning for this is that it allows further extension, 
like matching against attributes or body text.

For example, the DigesterContext implementation could be made pluggable 
to allow real document object models that enable full-blown XPath 
pattern matching. Of course that's an extreme example, what I want to 
express is that if the API is changed, it should be done with further 
extensions in mind.

I've posted my PROPOSAL a couple of minutes ago, maybe you wanna read it 
(but it's looooooooong ;o))

Christopher Lenz
/=/ cmlenz at

To unsubscribe, e-mail:   <>
For additional commands, e-mail: <>

View raw message