ietf
[Top] [All Lists]

Re: DNS based URI without any set access semantics?

2002-07-25 16:23:28
On Thu, Jul 25, 2002 at 06:28:25PM -0400, Keith Moore wrote:
| > | the year is not precise enough as domain ownership doesn't change on 
| > | year boundaries, and domain ownership can change more than once in a
| > | year.
| > 
| > Ok.  How about just YYYY iff YYYYMMDD would be YYYY0101
| 
| I don't like it;  it makes it harder to parse with NAPTR.

Ok.

| not if it's a URN; there are restrictions on how URNs are to be used.

Ok.  Perhaps I'm barking up the wrong tree

On Thu, Jul 25, 2002 at 03:45:16PM -0700, Ted Hardie wrote:
| > I'm just after a simple thing; a DNS based identifier that
| > isn't associated with a given protocol.  How people use the
| > identifier is up to them.
|
| And this is why you're seeing pushback.  Knowing what the tool is
| going to be used *for* gives you a shot at making a good tool.  Making
| a tool out of a certain substance without knowing how the tool will be
| used may very well produce sub-optimal results.  A glass hammer may be
| asthetically pleasing, but it probably won't be too good when you're
| ready to pound your nails.
|
| It sounds like you really do have a serious use case:  identifiers for
| YAML.  Writing up that use case may very well help you or others
| develop a good tool.  It might even be the case that the tool references
| the DNS.  But starting with "uses the DNS" doesn't seem like the
| right road to reaching "good tool".
|

Thanks.  Focus is good.  Let's stick to YAML's specific requirements,
sorry for attempting to generalize too soon.

YAML is a data serialization language.  It serializes, in a language
independent manner objects from various languages, including but not
limited to java, python, ruby, perl, COM, etc.  As part of an object's
serialization, it saves the component parts (data) as well as the 
object's type (class/package/etc.) in an text format.

Thus, I'm looking for a clean way to uniquely write "type" names 
across a great deal of languages with many internal ways that they
express their own object type names.  Ideally, this would allow a 
for each language's typing to be represented in a recognizable manner.
Also, there are some "types" which seem to be universal across languages,
such as integers, strings, etc.

Perhaps someting like a two-level mechanism would work...
  
  pkg:java:com.clarkevans.package.MyClass   # Java's reverse DNS package
  pkg:perl:Data::Denter                     # Perl's CPAN package
  pkg:python:yaml.loader                    # Python's package (not DNS)
  pkg:*:integer                             # language independent

Something like this would be even better than a simple DNS
hierarchy; although it would be more work as each language
would have to be defined with a specification of what sort
of sub-package is allowable.   It would be _alot_ of work,
but I think that this is a critical component of YAML (and any
cross-language system) so I'm willing to put forth the effort. 

Thoughts?

Thank you all (Michael/Ted/Keith) for helping me thus far.

Clark
Yo! Check out YAML, http://yaml.org 
Serialization for the masses

-- 
Clark C. Evans                   Axista, Inc.
http://www.axista.com            800.926.5525
XCOLLA Collaborative Project Management Software



<Prev in Thread] Current Thread [Next in Thread>