Re: ID specification?

From: John Halleck (John.Halleck@utah.edu)
Date: Fri Jul 06 2001 - 18:23:00 CEST


On Fri, 6 Jul 2001, Richard Knapp wrote:

> [... Unique Record ID's ...]

 Very interesting discussion...

> >http://rubens.its.unimelb.edu.au/~pgm/uisic/exchange/exchprop.html
>
> Sounds like namespaces. I'll give it a read. Thanks for the reference.

  I read it. Well done, but one has only moved one set of problems into
  another. I don't like where it breaks down to assigned numbers for
  part...

  Misc Issues:

    Out this end of the world, due to (in my opinion) excessive secrecy,
    there are caves known by different names by different groups.
    So different groups are going to get different numbers for the same
    cave, and you are still going to have to deal with combining issues.

    Including the organization name can lead to political issues (at
    least here). Which of the organizations surveying the cave is
    the one who's name is used. I've seen fights here about exactly
    who's survey project a given cave is. Having to pick one is
    troublesome.

    Many times I've had to go in and rename things in our data because
    a point that was thought to be the same turned out to be different,
    or we later discovered them to be the same, or we find that two
    groups chose the same survey station prefex. I don't see this
    as easy to deal with in that formal manner.

    The scheme given requires coordination between all surveyors on a
    project.. This is always troublesome. It would be nice if everyone
    always cooperated, but it is not realistic.
  
  Uniqueness for a survey is easy in a number of manners.
  (Munged date/time and email address of main surveyor, to pick a bad example)
  The real question is what this is going to be used for.
  If it is combining systems, then any two schemes that are pairwise unique
  will work with some conventions as to how you say that (for example):
      "John.Halleck@utah.edu:LBCC:19821112WSA1"
  is the same point as:
      "nahaj@u.cc.utah.edu:LittleBrush:19850914:wba1"
  The above example is intentionally ugly. Usable unique and large
  seem to me to be mutually exclusive.

  Personally, and I expect that others will have other views, I would lean
  towards an "implied prefix" tag, and simple locally unique names. (And
  staying away from XML ID's alltogether.)
  That way when things are combined I only have to **** with the implied
  prefix tag at the start of a section, and anything that says equivalence,
   instead of making changes to everything in the section.

> You've raised some interesting points and more questions.

  Very interesting points.

> [...]



This archive was generated by hypermail 2b30 : Mon Sep 03 2001 - 06:00:00 CEST