<html>
<head>
<meta content="text/html; charset=UTF-8" http-equiv="Content-Type">
</head>
<body text="#000000" bgcolor="#FFFFFF">
<div class="moz-cite-prefix">Hi Omar<br>
On 15/12/2015 19:08, Laurino, Omar wrote:<br>
</div>
<blockquote
cite="mid:CAFwvi4tKwzmhF7j97PhmXMoQuKnkeYO0byVz5NJBHB9ExWmYdQ@mail.gmail.com"
type="cite">
<meta http-equiv="Context-Type" content="text/html; charset=UTF-8">
<div dir="ltr">
<div class="gmail_extra">Francois,</div>
<div class="gmail_extra"><br>
<div class="gmail_quote">On Tue, Dec 15, 2015 at 11:55 AM,
François Bonnarel <span dir="ltr"><<a
moz-do-not-send="true"
href="mailto:francois.bonnarel@astro.unistra.fr"
target="_blank">francois.bonnarel@astro.unistra.fr</a>></span>
wrote:<br>
<blockquote class="gmail_quote">PS : I am not considering
here any kind of serialisation of these VO-DML
decscriptions (Object relational maaping, VOTABLE Mapping,
specific-stc2-xml-schema xml documents, json, etc... ).
this is another story but the initial VO-DML description
will constraint the solution. </blockquote>
</div>
<br>
I mostly agree. However, one of the benefits of VO-DML is to
have a language that can easily be mapped to a variety of
contexts, including OO languages and relational databases, and
so ORM. If variable length arrays were impossible (or very
hard) to treat in ORM, then they should be left off the
language, because they would be impossible (or very hard) to
treat in a fundamental set of implementations.</div>
<div class="gmail_extra"><br>
</div>
<div class="gmail_extra">But yes, the implementation details of
how to map VO-DML model instances to a serialization is a
different topic. So for instance, a model might call for a
great number of Object Type instances, but specific
serialization or representation formats might allow more
efficiency, whether we are talking about Java objects or SQL
tables, or FITS. An image might have, in principle, an Object
Type representation for each pixel in an image, but a FITS
representation might factor out all the common metadata into a
single representation for all the pixel in the image.</div>
</div>
</blockquote>
I think I agree with all that<br>
<blockquote
cite="mid:CAFwvi4tKwzmhF7j97PhmXMoQuKnkeYO0byVz5NJBHB9ExWmYdQ@mail.gmail.com"
type="cite">
<div dir="ltr">
<div class="gmail_extra"> I say in principle because I am not
sure I can take a side in the modeling part per se, so I don't
have an opinion on whether a pixel should be an Object Type or
a Data Type, in this example.</div>
</div>
</blockquote>
It is really strange for me to have coefficients as independant
Objects (instances of ObjectType). <br>
- Semantically a coefficient is basically just a number and
become a coefficient only in the context of polynomial. And we want
anything done for polynomials in the STC model to be independant of
the order<br>
- As both Arnold and Laurent stated in different wordings one
of the usage of UML (and attached VO-DML description) is generation
of code. ObjectType will become classes and having instances of that
for "coefficient" and "pixel" will lead to considerable overhead.<br>
<br>
Cheers<br>
François <br>
<blockquote
cite="mid:CAFwvi4tKwzmhF7j97PhmXMoQuKnkeYO0byVz5NJBHB9ExWmYdQ@mail.gmail.com"
type="cite">
<div dir="ltr">
<div class="gmail_extra"><br>
</div>
<div class="gmail_extra">Omar.</div>
<div class="gmail_extra"><br>
<div><br>
</div>
-- <br>
<div class="gmail_signature">
<div dir="ltr">Omar Laurino<br>
Smithsonian Astrophysical Observatory<br>
Harvard-Smithsonian Center for Astrophysics
<div>100 Acorn Park Dr. R-377 MS-81</div>
<div>02140 Cambridge, MA<br>
<span><a moz-do-not-send="true" value="+16174957227">(617)
495-7227</a></span></div>
</div>
</div>
</div>
</div>
</blockquote>
<br>
</body>
</html>