I'm fleshing out the tests for SOLR-4196 and am about to tackle "share
schema" tests. It strikes me that this is no longer a valid
distinction and want people to weigh in if they have strong feelings.

Currently, "shareSchema" is defined at the <cores> tag. Presumably
when individual cores specify a schema that points to the same schema
as another core _and_ shareSchema=true, the same underlying instance
is used. There's even a test for this. Otherwise, even though they may
be the same file, they're different objects.

In the new solr.xml-less world, I'm thinking the right thing is to do
the following (straw-man).
1> deprecate shareSchema and shareConfig in the <cores> tag (or
properties equivalent).
2> deprecate "config" and "schema" properties of an individual core
3> add an optional "configset" as an individual core property.
4> automatically share all identically-named configsets across cores.
4a> if no configset was specified in an individual core, use the
defaults found in the directory.

I don't see the use-case in allowing both "configset" and "config" or
"schema" in each core. This just seems too complicated to either
understand or get right. That said, I haven't really delved into all
the nooks and crannies that this would imply.

I'm thinking of opening up another JIRA for this bit, 4196 is large
enough already.

---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]

Reply via email to