Mailing-List: contact dev-help@asterixdb.incubator.apache.org; run by ezmlm
Precedence: bulk
Reply-To: dev@asterixdb.incubator.apache.org
MIME-Version: 1.0
Sender: ceej@lambda.nu
In-Reply-To: 
 <CALgBV_cMEvqnpO2ijdh2tKdoyeJR3T_TXeXWmJqKDCJor6=voQ@mail.gmail.com>
References: <55BCF130.1090805@gmail.com>
 <5F1DCABB-A9D4-4EF1-88B4-85AE6053F40B@apache.org>
 <55C1B59C.6040400@gmail.com>
 <CAHWyyRwnX1szvLHN12v4-udx1WpKZZrBnk5ypy-hgS5iTLPAkg@mail.gmail.com>
 <CAEYY6daLrvnLDfFBbqvVqo6oee15YXQ0SJFdFH7jg+Zki2j9yA@mail.gmail.com>
 <CAHWyyRzyBmJXE13LQjJLGt1t=13uMe4=G4XKCZ_qzrjtxLmyDg@mail.gmail.com>
 <CALgBV_cMEvqnpO2ijdh2tKdoyeJR3T_TXeXWmJqKDCJor6=voQ@mail.gmail.com>
From: Chris Hillery <chillery@hillery.land>
Date: Wed, 5 Aug 2015 16:10:10 -0700
Message-ID: 
 <CAHWyyRy-i+n9n5nX+f2YqoeqjhZS8tApTPkopmzvCPZEkVaKng@mail.gmail.com>
Subject: Re: json vs. JSON
To: dev@asterixdb.incubator.apache.org
Content-Type: multipart/alternative; boundary=089e0158b8e848bf65051c98820c

--089e0158b8e848bf65051c98820c
Content-Type: text/plain; charset=UTF-8

On Wed, Aug 5, 2015 at 12:23 PM, Wail Alkowaileet <wael.y.k@gmail.com>
wrote:

> 1- Line with *start* and *end* attributes give it a sense of "direction"
> which means {"start": [x1,y1], "end": [x2,y2]} != {"start": [x2,y2], "end":
> [x1,y1]}.
>

I take it that you think that's not correct? Is the ADM "line" explicitly
non-directional?

I had the same concern, honestly. As I said, my proposal was in part a
strawman, and this is exactly the kind of commentary I needed. Not clear
what better names would be... maybe just "a" and "b"?


> 2-Similarly for the rectangles, it seem there are some direction on the
> lines forming the rectangle.
>

Ditto the above. Just "a" and "b"? I was going to go with something like
"upper-left" and "lower-right", but that would require some logic in the
serializer to determine which was which that I don't think is appropriate.
It also wouldn't allow for a rectangle to be defined by the other two
corners, which I assume ADM is OK with.


> * I would suggest something similar to JTS (Java Topology Suites) where
> there are some generalization like for instance *rectangle *is-a *polygon
> *with
> two-points.


I think - kind of like GeoJSON (see next email) - that this suggestion is a
bit out of scope. That would be a good discussion to have regarding the
actual definition of the ADM types themselves, but it doesn't make sense
for the JSON serialization to somehow coerce the ADM meaning into some
other form. The serialization should be a very thin veneer over the actual
type definition.

Ceej
aka Chris Hillery

--089e0158b8e848bf65051c98820c--