SODA erratum 3 proposal

[Patrick Dowler]

I think the example in 4.2 (and the one in appendix A) are still consistent
with the decision we made and the relevant change is to define how to do
min/max for intervals (in DALI). Maybe that's an erratum on DALI, or just a
clarification in the current WD...

I feel like the change in wording in WD-VOTable means that DALI (where
xtype="interval" is defined) could (could have) defined min/max behaviour
to be more like that pointed out by Adrian in the PyVO context. That is:
use a 2D array value for MAX to specify the minimum bounding interval.

In general, when using an nD array (n matching xtype expectations) only the
MAX is really useful because the "minimum" is inside the maximum. The only
bit of info not conveyed there is the "minimum useful size of a cutout",
which in a basic non-complex extraction is the size of 1 pixel. That's true
for all axes  (see below).

So in my opinion MIN/MAX scalars like we already have, a MAX array (bounds)
would work, WD-VOTable defers to the xtype-definition but seems to allow
both (in principle), and DALI needs to be clarified.

my 2c,

below: yeah, a more sophisticated cutout implementation could do some
sub-pixel magic to figure out a fraction of signal to keep in the output
data, but I don't think we're talking about that level of detail
--
Patrick Dowler
Canadian Astronomy Data Centre
Victoria, BC, Canada


On Fri, 3 Feb 2023 at 08:30, BONNAREL FRANCOIS <
francois.bonnarel at astro.unistra.fr> wrote:

> Dear DAL followers,
>
>      I wrote this erratum proposal ten days ago, following an old issue
> raised first by Alberto Micol.
>      At the same time there was a merged PR in VOTable
> (https://github.com/ivoa-std/VOTable/pull/30)
>
>       which apparently discards the need for the SODA erratum.
>
>        See the discussion between Markus and me in VOTable issue 32v
> (https://github.com/ivoa-std/VOTable/issues/32)
>
>        So Markus suggests that for BAND interval MIN is the 1D array
> minimal value of both the lower and higher bound of the interval
>        and that MAX is the maximum value of the same two bounds.
>
>        In such a  way that :
>                     1 ) a 2D array MIN and MAX value attribute is not
> required
>                     2 ) we forget the idea to have MIN as the minimal
> interval and MAX the maximum interval
>          However interval is an xtype so we could have specific rules in
> this case as stated by the new VOTable text. So ....
>         Thoughts ?
>
>         IN any case if we follow Markus, we still have to fix the wrong
> tag </VALUE> and clarify the example meaning following the new text in
> VOTable
> So there will still be an erratum 3 on the same example
> Cheers
> FRançois
> Le 23/01/2023 à 15:04, BONNAREL FRANCOIS a écrit :
> > Dear all,
> >
> >     This is a proposal for an erratum of the SODA 1.0 specification.
> >
> > https://wiki.ivoa.net/twiki/bin/view/IVOA/SODA10Err3
> >
> >   Please check it and let's start the discussion.
> >    I wonder if this is an erratum or a real change (for SODA 1.1).
> >
> >    I looked at VOTable, and although it seems reasonable, if I am not
> > ignoring some rather hidden text or xsd, nothing seems to require that
> > arraysize should follow the PARAM or FIELD arraysize.
> >    I addition if we do this change there is another option (as
> > mentioned in the associated github issue) related to the meaning of
> > these MIN and MAX arrays. Are the two values in MIN (resp MAX)
> > defining the minimum (resp maximum) length interval (which I find
> > consistent with the definition of BAND in that case) or the range of
> > possible values for the lower (resp higher) limits of the BAND interval.
> >
> > Cheers
> >
> > François
> >
> >
> > Le 23/01/2023 à 07:46, BONNAREL FRANCOIS a écrit :
> >>
> >>
> >> Dear all,
> >>
> >>     This is a proposal for an erratum of the SODA 1.0 specification.
> >>
> >> https://wiki.ivoa.net/twiki/bin/view/IVOA/SODA10Err2
> >>
> >>   Please check it for fast acceptance.
> >>
> >> Cheers
> >>
> >> François
> >>
> >
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.ivoa.net/pipermail/dal/attachments/20230213/c3d6fecb/attachment.htm>