Very Computer

   // IXMLDOMNode* attribute_node;
   // VARIANT varAttributeValue;
   attribute_node->get_nodeValue(&varAttributeValue);
   BSTR bstrAttributeValue = V_BSTR(&varAttributeValue);

Even if the original XML file has either UTF-8 or ISO-8859-1 encoding, and
contains diacretic symbols encoded as entities ("é"), this BSTR will
contain these entities as "normal" ASCII values. Of course, one could
translate these values back by parsing the attribute value, but this would
cause a performance hit *twice*.

Is there a more elegant solution to this? Would this require _bstr_t, or a
stream with a specified encoding or such? I'm really looking forward to know
about this.

Regards,

Toine de Greef

Quote:> I'm using msxml3 for instance like this:

>    // IXMLDOMNode* attribute_node;
>    // VARIANT varAttributeValue;
>    attribute_node->get_nodeValue(&varAttributeValue);
>    BSTR bstrAttributeValue = V_BSTR(&varAttributeValue);

> Even if the original XML file has either UTF-8 or ISO-8859-1 encoding, and
> contains diacretic symbols encoded as entities ("é"), this BSTR will
> contain these entities as "normal" ASCII values. Of course, one could

Quote:> translate these values back by parsing the attribute value, but this would
> cause a performance hit *twice*.

Quote:> Is there a more elegant solution to this? Would this require _bstr_t, or a
> stream with a specified encoding or such? I'm really looking forward to
know
> about this.

> Regards,

> Toine de Greef

I think you helped me by asking the right question: "Why would you want to
translate them back?". I was going to answer: "Because after doing some
other stuff with it, no XSL(T), I want to write it back into an XML
document." And I thought that other document would dislike these literal
diacretical entries (being UTF-8). Everything will work fine though., when
using ISO-8859-1 for both documents.

Thanks again,

Toine.

Quote:> Julian,

> I think you helped me by asking the right question: "Why would you want to
> translate them back?". I was going to answer: "Because after doing some
> other stuff with it, no XSL(T), I want to write it back into an XML
> document." And I thought that other document would dislike these literal
> diacretical entries (being UTF-8). Everything will work fine though., when
> using ISO-8859-1 for both documents.

As a workaround, I used entity notation for them. When using
get_nodeValue(), to my surprise these values where translated, thus breaking
my workaround. I even tried setting encoding="ISO-8859-1", but that did not
change the behaviour of  get_nodeValue().

My conclusion: no bug.

Quote:> Hopefully we're not stretching this out of proportion, but what I was
saying
> is te following. ISO-8859-1 diacretic symbols not written as entities
(i.e.
> "?"), in an XML file with UTF-8 encoding, violated encoding constraints. I
> assume this is in accorandance with XML processor requirements.

Quote:> As a workaround, I used entity notation for them. When using

[...]

Quote:> Why not just use UTF-8 encoding, if this is what the file is declared to
> use?

? in ISO-8859-1: ?
? as entity reference: ë or ë or ë
? in UTF-8: ??