Re: XmlSerializer

>>when the XmlSerializer is writing to a reader with unspecifyed encoding,
>>I think we should require escaping all character codes > 127? Is that OK?
>>
> 
> you mean to escape using XML entity character references (Ҽ ) right?

yes.

> i do not know - shouldn't it be left to implementation? default XML encoding is
> typically UTF8 and it can transfer any unicode character without need to
> do XML entity escaping.

yes, but if a reader is given and no encoding, we cannot expect utf8,
I think we should really follow a "safe" path here...

> actually another hint to serializer is
>     startDocument (String encoding, Boolean standalone) throws IOException;
> where user can declare even for Writer what is intended XML encoding.
> 
> so i think that requiring it is not needed we should instead require
> that every implementation takes into account encodings
> passed in setOutput() and startDocument().
> should we require both encodings to be identical?

yes we should require that they match.

I still think when no encoding is given, we should assume
no more than 7bit ASCII: When writing using a platform
default writer, the encoding will probably be ISO-Western,
but by using just 7 bit we can make sure
that a parser does not run into trouble when infering utf8 from
the lack of the encoding declaration.

Best,
Stefan