sexagesimal

[Rob Seaman]

Roy wrote:

> The question is how to recognize that a string is actually  
> sexagesimal coordinates. If a text box on a form allows coordinates  
> or object name, how can code be written to decide what it is.

Our NVO-SS contingent has returned and has been able to report on the  
context of this discussion.  It might help if folks would state the  
problem fully in advance of suggesting a solution :–)  Not all of us  
are present at every VO face-to-face.

> The answer I am seeing is that you spend 3 days writing 500 lines  
> of Perl to disambiguate the numerous symbols and endless lemmas and  
> codicils, then you fight like hell to keep that code useful by  
> refusing to constrain the sexagesimal format even in the slightest,  
> and forcing every other developer to go through the same 3 day  
> bootcamp that you did.

Well, yes, if there is some requirement to support all known  
variations of sexagesimal notation - especially if users are being  
encouraged to enter both RA and Dec into the same text box for some  
reason.  I'm skeptical that there is such a requirement beyond  
importing legacy catalogs from ASCII sources.

Heck, I'm happy with my colon...er, that didn't come out right...at  
any rate, there's no reason to support all formats. Like I said in a  
previous message, the heuristic you described (three numbers  
separated by two non-numbers) is ok for government work.  The part  
you're missing is code to detect ambiguous specifications.  The  
solution isn't to try to legislate human data entry, but to detect  
situations that cannot be reliably parsed into a single clear value.   
Then report these back to the user for correction.

It seems to me that the real problem is with the assumption that a  
single box on a form should be prepared to accept both names OR pairs  
of coordinates.  Instead of a single text entry box accepting either  
<name to be resolved> or <coordinate pair to be resolved>, how about  
three boxes that the user can select between depending on what they  
have at hand.  See http://archive.noao.edu/nsa for an example.  It  
isn't the sexagesimal notation that needs to be constrained, it's the  
idea that the user should be encouraged to type any random damn  
quantity he wants into some form, rather than the form supplying a  
separate entry field for each semantically separate type of data.

Steve said:

> I don't know about Rob, but before I was an undergraduate I was  
> deriving the hyperbolic geometry equivalents on little notepads at  
> the nurses station in between changing bedpans during the hospital  
> night shift.

...and the question is whether the expertise with hyperbolic geometry  
or bedpans will prove move useful in the IVOA era :–)

Rob
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.ivoa.net/pipermail/dm/attachments/20060918/87b0ba62/attachment-0001.html>