URI fragments in IVOA specifications

[Norman Gray]

Dick and all, hello.

I've taken a while to reply here -- apologies.

I've put another version of this draft note at <http://www.astro.gla.ac.uk/users/norman/temp/uri-fragments.html>

On 2012 Apr 5, at 18:58, Dick Shaw wrote:

> 1. Though I understand your original motivation, to correct what you see as a problem with how URI fragments are used in practice, I think I would re-cast the content of your document more positively. That is, make the emphasis on best practices for usage of URIs in the IVOA context. Certainly there is a place for discussing caveats, but many potential readers will benefit more by seeing examples of proper usage than be being admonished about improper usage.

I haven't followed your advice here, because I'm sorely tempted to!

I think such a document -- giving advice on good practice for URLs, and the like -- would be interesting and valuable, and I've included a 'to-do' note within the document wondering aloud whether it should be expanded to cover this (and I wouldn't take much persuading to start on it).  However I can imagine that such a document would turn into a much longer thing, much debated, and full of people riding hobbyhorses (I have a few hobbyhorses on this topic which I keep permanently saddled up).  That would be entertaining, but less focused than the current draft note.

> 2. Consider making the content more approachable to those with lesser knowledge of the standards. The current text has very much the flavor of proving that certain usage may be wrong. That's not exactly a flaw, but the current reference list of supporting material is highly technical, and oriented to the expert. Perhaps this is why bad usage has cropped up in the first place. An introductory paragraph explaining the basic concepts (and perhaps references to tutorial material, if it exists) would be helpful for understanding the basics of URIs.

I've done this.  I think that a problem with the issue is that it is a consequence of a subtlety in the mandated behaviour of URIs, which is generally under-appreciated, and which I (and I'm sure I'm a finicky constitutional lawyer in a parallel life) had never properly thought about before reading one of the final drafts of the VOEvent spec and thinking 'that smells funny'.

That said, there's no reason to have it more opaque than necessary, and I've added a couple of paragraphs of prefatory text which I hope explain the issue in broad terms.  I'd be very interested in whether you agree that what's been added is useful or sufficient.

> 3. Provide examples within IVOA where the URI usage is both proper and illustrative of best practices (I assume there are some).

I've added a brief paragraph on this, too.

Best wishes,

Norman


-- 
Norman Gray  :  http://nxg.me.uk
SUPA School of Physics and Astronomy, University of Glasgow, UK