<div dir="ltr">Dear all,<div>the idea behind the XMATCH function was a simplification over the CONTAINS solution, as detailed by Mark and Markus, to have an easier way for the user to have that [maybe too simplified] crossmatch.</div><div><br></div><div>I'm happy that the post to define point-or-coordinates raised higher level requests for real cross-matching capabilities in ADQL (currently in TAP services), but I don't think this (i.e. ADQL-2.1 minor revision) is the place to bring in such discussion, consider the XMATCH function itself will be only optional.</div><div><br></div><div>That said, I suppose we have to face a fork:</div><div>- we continue with the proposed XMATCH and discuss on point vs coordinates</div><div>- we drop XMATCH from ADQL-2.1 and keep on with a general discussion for a later major ADQL revision</div><div><br></div><div>My vote goes for the first one, since it seemed to be quite an easy (even if incomplete) answer to use cases. But I'll not protest if the other direction is taken.</div><div><div class="gmail_extra"><br></div><div class="gmail_extra">I also agree with Laurent's point on optional capabilities of single services and the need for specific optimization in that case (his mail dropped in while I was writing).</div><div class="gmail_extra"><br></div><div class="gmail_extra">Cheers,</div><div class="gmail_extra"> Marco</div><div class="gmail_extra"><br></div><div class="gmail_extra">PS - for the second option above I kindly suggest to link something in <a href="http://wiki.ivoa.net/twiki/bin/view/IVOA/ADQL">http://wiki.ivoa.net/twiki/bin/view/IVOA/ADQL</a> or <a href="http://wiki.ivoa.net/twiki/bin/view/IVOA/ADQL-2_0-Next">http://wiki.ivoa.net/twiki/bin/view/IVOA/ADQL-2_0-Next</a> in the meanwhile ADQL-2.1 goes REC, after that ADQL-2_1-Next will host the proposed features.<br></div><div class="gmail_extra"><br><div class="gmail_quote">2016-02-10 9:46 GMT+01:00 Alex Szalay <span dir="ltr"><<a href="mailto:szalay@jhu.edu" target="_blank">szalay@jhu.edu</a>></span>:<br><blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left-width:1px;border-left-color:rgb(204,204,204);border-left-style:solid;padding-left:1ex">Then we should talk about how to simplify and minimize, but to implement something that is<br>
Both conceptually and mathematically incorrect would be a big mistake.<br>
<br>
I am travelling in New Zealand for the next ten days, and today I happened to have a good<br>
internet, but mostly I am off-line. So this is not an ideal time for me to participate in an<br>
intense discussion.<br>
<br>
I am also cc'ing Laszlo and Tamas on the trail, in any case<br>
<span class=""><font color="#888888"><br>
--Alex<br>
</font></span><span class="im"><br>
<br>
-----Original Message-----<br>
From: <a href="mailto:dal-bounces@ivoa.net">dal-bounces@ivoa.net</a> [mailto:<a href="mailto:dal-bounces@ivoa.net">dal-bounces@ivoa.net</a>] On Behalf Of Walter Landry<br>
Sent: Wednesday, February 10, 2016 3:37 AM<br>
To: <a href="mailto:dal@ivoa.net">dal@ivoa.net</a><br>
Subject: Re: ADQL XMATCH<br>
<br>
</span><div class=""><div class="h5">Alex Szalay <<a href="mailto:szalay@jhu.edu">szalay@jhu.edu</a>> wrote:<br>
> This is a paper we published 4 years ago about this. The service is<br>
> now in beta testing. We just bought 4 dedicated servers.<br>
><br>
> <a href="http://arxiv.org/pdf/1206.5021v1.pdf" rel="noreferrer" target="_blank">http://arxiv.org/pdf/1206.5021v1.pdf</a><br>
<br>
This is interesting work. However, I would wager that this would be overkill for most of our users, who would be fine with a simple closest match. For the rest, it would be insufficient, because they want to use error ellipses or filters on object type<br>
<br>
<a href="http://arxiv.org/abs/1503.01184v1" rel="noreferrer" target="_blank">http://arxiv.org/abs/1503.01184v1</a><br>
<br>
So I would be wary of standardizing anything too sophisticated.<br>
<br>
Cheers,<br>
Walter Landry<br>
</div></div></blockquote></div><br></div></div></div>