Tobi_WMDE_SW added a subscriber: Tobi_WMDE_SW.
Tobi_WMDE_SW added a comment.
@daniel is this resolved by now?
TASK DETAIL
https://phabricator.wikimedia.org/T118860
EMAIL PREFERENCES
https://phabricator.wikimedia.org/settings/panel/emailpreferences/
To: daniel, Tobi_WMDE_SW
Cc:
adrianheine added a comment.
Not quite. I want to see facet usage to have more opportunities at explaining
you why I think they are bad, since my abstract points didn't get to you yet :)
TASK DETAIL
https://phabricator.wikimedia.org/T118860
EMAIL PREFERENCES
daniel added a comment.
@Tobi_WMDE_SW *sigh* no. I need to write down an updated plan, to summarize the
current state of discussion. I'm pushing back the final decision on the details
until we are have a concrete plan to work on this. Perhaps we can talk about
that at the next story time?
adrianheine added a comment.
I can't really debunk this idea without having any code that uses it. The PR in
data model doesn't look too bad, just unnecessary, but I'm pretty sure I'll
have something against code that would use it. Maybe you try the RDF thing with
this and I and others can
daniel added a comment.
Summary of today's discussion with @thiemowmde:
- Don't replicate code in the model, use a common base class (or trait), so
facets are allowed in all data model nodes.
- Inject facet declarations (a FacetDefinition service or some such) into each
object that supports
daniel added a comment.
Summary of some a brief discussion with Adrian and Jan:
With the Role Object approach, consumers of extra data cannot declare that they
want to use that data. With lookup/mapping services, consumers can declare the
dependency by requiring the respective services to be
