MANGO DM in working draft

Laurent Michel laurent.michel at astro.unistra.fr
Thu Jan 30 17:19:54 CET 2025


Hello DM,

Than you again Mark.

Mango have been built after a poll in the community to get use cases for a source model. We got then many contributions and many use-cases. I agree that MANGO does not support all quantities mentioned in the use cases yet. We need probably a small section telling how the selection has been done.

Basically MANGO is a container for properties but nothing prevent it to provide its own classes and to mix them with imported classes.   

> On 29 Jan 2025, at 23:31, CresitelloDittmar, Mark <mdittmar at cfa.harvard.edu> wrote:
> 
> Use Cases/Requirements
> Between this section and the Appendix, there is a lot of detail on use cases.  I'd like to see more clearly:
> the assessment of the data to be associated by the MANGO model.
>  ie: what kinds of data:  Properties ( Meas, Flags, Assigned .. );  Data Products;  <other?>
> and which are selected to be supported by this version of the model.
> There are specific properties listed in the cases, and also broad statements like "approximately 1,700 columns covering properties", and "set of stellar host characteristics".  Clearly, MANGO isn't going to include the specs for all of these Properties.  So seeing the assessment of the Types of Properties to be supported, and an example from each (defined either internally or externally but in such a way that they appear external) seems important.
> 
> Requirements:
> I think that several of the requirements on MANGO are not actually MANGO requirements, though I think some of this is a difference in opinion about MANGO’s scope.  MANGO should have requirements to support the various flavors of property, but not the definition of those properties

yes it does (see above)
> eg: MANGO must support Properties of measured quantities, including associated coordinate systems and errors but it is not required "to attach relevant coordinate systems to any measured quantity"

Mango does not tell how measured properties have to be done. The internal property structure is part of the property definition (imported or not).
> eg: "MANGO must <snip> properties which attributes spread out on multiple columns (eg positions, errors)";  The Property (Measure) handles the multi-column associations, MANGO is merely responsible for connecting to the Measure.
> eg: "MANGO must <snip> provide an accurate description of the epoch propagation. <snip> consists in constructing 6 parameter position vectors…"; 
> The epoch propagation usage thread was selected for implementation, so it could be a requirement on MANGO to support the properties and associations necessary to fulfill that case.  But MANGO is not required to construct a 6 parameter position vector.. There are multiple ways to solve that problem.
The EP is a GAIA requirement. It has been implemented in a way that matches the implementer expectations. 
I also exercised others options that aggregated structured components. This was perhaps more consistent but not usable in practice (difficult toi build, difficult to read).

> I think that, for the most part, the Requirements for the Properties themselves would fall to the standards for that type of property. 

I’ve that feeling as you can guess.
LM

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.ivoa.net/pipermail/dm/attachments/20250130/c8e13908/attachment.htm>


More information about the dm mailing list