[
Lists Home |
Date Index |
Thread Index
]
> -----Original Message-----
> From: Bullard, Claude L (Len) [mailto:len.bullard@intergraph.com]
> Sent: Friday, April 08, 2005 17:31
> To: 'Alessandro Triglia'; 'Elliotte Rusty Harold'
> Cc: 'xml-dev'
> Subject: RE: [xml-dev] The Rising Sun: How XML Binary
> Restored the Fortune s of Innovators
>
>
> Because the question you are asking is, will this make
> everyone happy,
> and I can only answer, it should with respect to ubiquitous
> view source
> given losslessness. I added the rest to point out that is only part
> of the consideration for the binary because there is a significant
> set of people who don't want their code to be view sourced,
> and performance
> is one concern. Yes, these are separable issues.
>
> It answers the view source issue as long as the viewers in a
> sufficiently
> large number of cases produce precisely the same XML display.
> This goes
> to the 'lossless' issue, I suspect, and to the 'provable code' issue.
>
I would like to make the assumption that many **producers** of XML care only
about the "infoset" portion of the XML they produce. In other words, they:
- don't need the ability to use general entity references
- don't need the freedom to choose the nature and amount of whitespace
inside tags
- don't need the freedom to choose between a starttag/endtag pair or an
empty element tag
- don't need the freedom to choose between apostrophe and quote for
attributes
- don't need the freedom to choose the order of attributes
- have no special desire to violate namespace-well-formedness
and so on.
I am not asserting that all producers of XML fit the above description, but
I like thinking that a lot of them do. Otherwise, it would be hard to
understand the significance of the XML infoset, the significance of SAX, the
significance of XML Schema, etc. The XML infoset is important, isn't it?
Now, if a producer of XML that fits the above description switches to
producing a fast infoset document instead of an XML document, all the
information that matters to them will be faithfully preserved in the fast
infoset document. Information that does not matter to them (such as
apostrophes/quotes, use of numeric character references, nature and amount
of whitespace in tags) may be lost. There is no loss of useful information.
One has a loss of some sort when one starts from an XML document and
converts it to FI and then back to XML. But if one starts from an XML
infoset, there is no loss.
In such cases, I think that a FI viewer that renders the fast infoset
document as XML 1.0 would satisfy the "view source" requirement.
Alessandro
> len
>
> From: Alessandro Triglia [mailto:sandro@mclink.it]
>
> > I don't think view source is the whole issue, but it is a
> > significant one to a
> > large and equally important community: authors/programmers.
>
>
> That does not answer my question.
>
> My question was whether the "view source" requirement would
> be satisfied if
> Fast Infoset viewers became ubiquitous.
>
> If some author/programmer is currently using the MS XML 1.0
> viewer built
> into IE as their "source viewer" (as opposed to using Notepad), what
> difference would it make to him if he had to use a FI viewer
> instead of the
> MS XML 1.0 viewer?
>
> (I know one of the possible answers - If the document is
> broken, the guy
> will have to use Notepad! This is not an answer that would
> satisfy me,
> though.)
>
|
