Re[2]: [cgmo-webcgm] ISSUE: delimited string inconsistency?

[Lofton Henderson <lofton@rockynet.com>]

At 09:53 AM 5/6/2005 -0400, Benoit Bezaire wrote:
>Hi Lofton,
>
>So are you still convinced that a single string is better than a
>delimited string for the 'region' attribute? I am not.

Okay, first lets clarify.  If you were going to apply delimited string to 
this rectangular region of a single subregion, are you talking about:

"'1 '0' '0' '100' '100'"  or  "'1 0 0 100 100'"   ?

I'm guessing the latter (the first is just a silly waste of quotes), so 
that a rectangular annulus would be:

" '1 0 0 100 100'  '1 25 25 75 75'  "

I.e., each sub-region is delimited.


>It seems like we would be re-inventing a single string syntax for this
>particular attribute when we've define a syntax to handle these kinds
>of problems (i.e., delimited strings).

In my view, delimited strings are necessary to handle the "strings of 
sub-strings" case, where the sub-strings can contain all sorts of stuff 
like blanks, commas, etc.

Region is a list of numbers.  A number is a consecutive string consisting 
only of digits, +/- signs, decimal point (and maybe "E", in the case that 
we allow scientific notation, -0.6e32 -- haven't thought about this one yet.)

I don't really see what "delimited string" adds to 'region', other than 
complexity of syntax.  By far the most common region will be rectangle, 
hence this region:  1,0,0,100,100

Do we really want,
"'1 0 0 100 100'"  (or "'1' '0' '100' '100'")?

To me, even a multi-sub-region is dead simple to handle:

" 1 0 0 100 100  1 25 25 75 75  "

I guess I could live with it either way, but don't see any reason to apply 
more powerful syntax to a simple problem like region.

Cheers,
-Lofton.