forrest-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "J.Pietschmann" <>
Subject Re: draft howto dtd
Date Sun, 19 May 2002 11:34:51 GMT
Steven Noels wrote:
> You live with the assumption standards are correctly implemented by
> tools, regardless of their complexity.

Not really.

> If a standard (or better: the
> application of a standard) can only live within a certain predefined set
> of tools, it isn't much worth as a standard.

Well, even many mainstream XML parsers don't implement
all obscure *DTD* features:
and there are rarely complaints about this.

I'll note that the number of features used in the currend
Apache DTDs is fairly limited, I haven't found, for example
any conditional DTD sections.

> My problem with XML Schema is that it offers too many possible ways of
> declaring the same kind of thing, and believe me, when you hand over
> such a language to a creative user, he will make use of the exotic
> features, too.

Using XSchema for defining structures of Apache documents
is not exactly the same as handing it to arbitrary creative
users. Nobody stops the initial committers from writing up
some guidelines to restrict the subset to use. This can even
be quite fuzzy, the community will self correct and amend the
guidelines as needed. For example if there are repeated
complaints that a certain feature won't work, there will be
a discussion whether it's really necessary to use it, and
either it will be dropped or replaced or the complaining
people are convinced and move to tools which support the
feature or work around the deficiencies. I firmly believe
that XSchemas for Apache documents will start fairly simple
and probably stay so for quite some time, introduction of
"exotic features" will be frowned upon unless they add
real value.
For a start, because XSchema is hardly universally supported
by editors, it would be necessary to generate DTDs from the
schemas, and of course documents which validate against such
a DTD and some verbal guidelines ("put only svg:* in this
ANY element") should validate against the schema. The
pressure to keep the additional verbal guidelines as small
as possible will prevent usage of most of the "exotic
features" for some time.

As for your particular example regarding elements with
default content: I'm in favor of the view that XSLT
processors should be able to do their work without being
forced to read DTDs or schemas. Declaring default values
there thwarts this, therefore I don't like it.

> And then, you can only hope your editor, validation
> engine et al. will implement the same kind of 'standard', which is
> currently not the case.

Well, XSchema is bloated and I think it will finally topple
under its own weight and be replaced by something else,
perhaps by some subsets or "profiles" defining validation
against certain sets of rules. Until then, we can stick to
DTDs and deal with the restrictions (no namespaces, ANY,
no processing by XSLT), or we can make up our own rules
on how to deal with the problems XSchema while taking
adantage of the virtues.


View raw message