schemaless configs should remove copyField to _text
or at least change _text name.
This default led to bad indexing performance.
Data driven solrconfig spellchecker refers to a non-existent field
schemaless example should have catchall field
The _text field and the copyField directive targetting it were added in SOLR-6471.
Anyone have thoughts on changing
"_text" to either "_text_" or just "text"
Personally I'd slightly prefer "text" to just "text", although I can argue either way. "text" is clearly a reserved field, OTOH "text" is something we already use as the df.
"_text" is a distant third IMO since it's too easy to confuse with a dynamic field.
I prefer _text_.
+1 for _text_ - any objections to changing this in 5.1? If we're going to change it, I think we should do it sooner rather than later.
+1 for text - any objections to changing this in 5.1? If we're going to change it, I think we should do it sooner rather than later.
+1 for changingin 5.1
I'm running tests now.
Added note to "Upgrading from 5.0" section in CHANGES.txt
Commit 1671234 from Steve Rowe in branch 'dev/trunk'
[ https://svn.apache.org/r1671234 ]
SOLR-7290: Rename catchall text field in data_driven_schema_configs to _text
Commit 1671235 from Steve Rowe in branch 'dev/branches/branch_5x'
[ https://svn.apache.org/r1671235 ]
SOLR-7290: Rename catchall text field in data_driven_schema_configs to _text (merged trunk r1671234)
Commit 1671236 from Steve Rowe in branch 'dev/branches/lucene_solr_5_1'
[ https://svn.apache.org/r1671236 ]
Committed the renaming patch to trunk, branch_5x and lucene_solr_5_1.
Note that this leaves the copyField behavior intact.
I'll resolve this issue - if people want to continue discussing the copyField behavior, please create another issue.
Bulk close after 5.1 release