Use Advanced Search to search the entire archive.
Re: Proposal
- From: Martin Desruisseaux <
>
- To:
- Subject: Re: Proposal
- Date: Sun, 02 Nov 2014 20:40:13 +0900
- Organization: Geomatys
Just in case the 3 listed points are the intended ones, there is my though:
Le 02/11/14 20:24, Martin Desruisseaux a écrit :
>
Remove the Measurement interface in order to avoid the conceptual
>
problem with hierarchy.
I would support that, since I believe that an interface named
Measurement is dangerous. We have see the difficulty to put it in a type
hierarchy. Furthermore the experts groups working on "Observation and
Measurements" international standards (I'm thinking about the Open
Geospatial Consortium here) could also point out that our Measurement is
very far to what an actual measurement is, since we carry no information
about measurement accuracy.
>
Overload the multiply and divide methods for type safety.
I'm neutral on this one. I like small API and avoiding multiplication of
methods doing almost the same things. But Jean-Marie has a point when he
said that the amount of methods would not necessarily be huge, since we
provide only about 40 standard Quantity sub-interfaces.
However this brings a new issue: if a sensor vendor wants to trim down
JSR-363 by retaining only the Quantity subtypes that the sensor
supports, removing the unused Quantity subtypes may not be sufficient:
he may also need to remove some of those overloaded multiply/divide
methods in the remaining interfaces.
Martin