Closed havardf closed 2 weeks ago
What about these rules for starters:
We can then add these to the profile now, and update later as needed. It would anyway be good to add something quickly, so it is easier for everyone later to contribute.
We should handle multiple languages. I think best way to do that is to have parameter such as lang=fi and then responses would be in that language. CoverageJSON does have something for the i18n, but that does not solve the whole problem.
Default would be English, which must always be available.
Title is now missing from many of the services. Should we say that it is mandatory, otherwise clients will default to collection id, which is maybe not optimal for user interfaces and leads to not optimal lists for the user.
I say definately yes. Maybe like this:
Mandatory and with the following rules:
There have been a lot of discussion internally in FMI about naming conventions, it is not easy, but one thing we observed was that names should have the most "important" bit first / near beginning of the string so if UI needs to truncate title, relevant part is not dropped. Your example follows this "Surface observations in Europe"
Prohibiting acronyms could cause problems when they are often more recognizable than what they originally stand for, e.g.
A title of "surface weather observations from NOAA" is not helpful to the actual content. Here, "NOAA" is a property of the entire record, as the organization.
I actually had a specific example in mind:
Which collection? I am guessing the collections with "UK" as part of the title (perhaps "Airfields in the United Kingdom" would work better here?).
Ignoring "UK" and "EDR", the following collections have titles with acronyms which would exceed the 50 char limit if expanded:
What value does "NOAA" have in an actual title? Can a user not derive this from the collection metadata itself?
Well, the only other relevant metadata field available under /edr/collections is description
, which is not always displayed unless reading the JSON manually (e.g. you can't see it when using Swagger UI).
I feel this is more of an issue for forecast data than for observations. When quering simulation data it is essential to know which model you are using, and these are more often than not named using acronyms (MEPS, AROME, ECMWF Hi-Res, UKMO, GFS, HRRR et al).
Is Swagger UI a requirement then? Note that there needs to be some provision around description
for fulsome, descriptive metadata.
Typical use-case: Client populates a selection menu with
title
of collections.Some potential rules: