[Mono-devel-list] Re: asorted xml implementation questions

[ginga at kit.hi-ho.ne.jp]

Hi Erik,

> 1) the more I look at their API I'm convinced that MS uses 
> derived classes of XmlSerializer and XmlSerializationWriter. 
> The problem is that I cannot see the advantage. 

Thank you for your informative analysis. Yeah, I think so too,
because both of them have some virtual methods. Then I thought
some advantages for us to use them ... e.g. XmlSerializer code is,
and may become huge monster, and they may be delegated into 
XmlSerializationWriter/Reader.

I think XmlSerializer have two aspects. One is a frontend for
xml serialization, and the other is such that is used as a base
class for custom, individual serializer. As the first aspect,
it have Serialize (XmlWriter, ..snip..) methods, and as the second,
it have Serialize (object, XmlSerializationWriter).

As I experimented, XmlSerializer doesn't call Serialize (object, XSW) 
in the first aspect, then it is likely to create another (custom) 
serializer and invokes its Serialize (object, XSW).

I thought that it might be good start for us to infer roles of 
those 'internally used' classes. I'll be grateful if you tell us
about any ideas for them. And on improving deserialization,
we will need to implement XmlValidatingReader for contributing
post schema validation information.

Thanks,
  Atsushi Eno