xml-xmlbeans-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From robert burrell donkin <rdon...@apache.org>
Subject Re: Future XMLBeans feature work?
Date Sat, 20 Sep 2003 09:20:18 GMT

On Thursday, September 18, 2003, at 09:22 PM, Chris Maeda wrote:

> Hi David,
>
> It seems to me that the "start-from-Java" and
> "fast lossy binding" features would be less valuable since
> (as you mention) there are other technologies out there
> to do this already, like Apache Axis.

IMHO start-from-java has quite a bit of hidden complexity. betwixt has 
been under development for a long time now but we're only really starting 
to understand the real questions.

i'd say that it would be a good idea to think about pluggable 
compatibility. i would whether it would be possible to use adapters to 
allow xml-bean generated, JAXB compliant and start-from-java mapped (eg 
using betwixt) to be used together.

(the rest may well be totally left field and if so, please forgive me...)

in a lot of ways, i think that one missing part of the jigsaw is an object 
pipeline to sit on top of the SAX (or SAX-like) marshalling layer.

it seems to me that a lot of the inefficiency (in typical usage senarios) 
now comes from building up large object models in memory. it's quite 
common to have large, repetitive xml documents where really, it's the 
smaller sub-elements only which want to be mapped to (and from) java. 
ideally, the objects for these sub-elements should be pooled. so, what you 
have is really a partial mapping of the xml graph to an object graph 
rather than a full one.

experience coders will undoubtedly work around the technologies to achieve 
this result but it seems to me that these technologies are now achieving 
greater penetration into the mainstream where coders may not be so 
familiar with the issues involved with mapping large documents. it'd be 
cool if the marshaling technologies could feed into an object pipeline 
that'd allow users easier and efficiently to handle big documents.

i'm a big fan of filtering but java coders are probably more familiar with 
filter request objects than SAX (or SAX-like) events. so, i'd say that it'
d be easier for most coder to understand filters and controllers in an 
object pipeline than a SAX (or SAX-like) one.

maybe good support for object pipelining might be another good way to xml 
beans to offer more than just simple marshalling.

- robert


- ---------------------------------------------------------------------
To unsubscribe, e-mail:   xmlbeans-dev-unsubscribe@xml.apache.org
For additional commands, e-mail: xmlbeans-dev-help@xml.apache.org
Apache XMLBeans Project -- URL: http://xml.apache.org/xmlbeans/


Mime
View raw message