SQL: do not ignore all fields whose names start with underscore #36214
Conversation
|
Pinging @elastic/es-search |
|
The test failure comes from the docs build. It installs the |
|
@nik9000 is the In case we need to keep it, @atorok do you know a way of running |
|
The mapper-size plugin is installed here because we run tests for its docs. We install almost all of the plugins when we're testing the docs. I don't think we should run the SQL tests in a cluster without the mapper-size plugin installed. I think SQL shouldn't include the |
| // Skip internal fields (name starting with underscore and its type reported by field_caps starts with underscore | ||
| // as well). A meta field named "_version", for example, has the type named "_version". | ||
| // Also, skip any of the blacklisted field names. | ||
| if (false == (name.startsWith("_") && entry.getValue().values().stream().anyMatch(f -> f.getType().startsWith("_"))) && |
There was a problem hiding this comment.
Potentially move the types declaration above the if to reduce the if:
types.values().stream().anyMatch()
Another option would be to do the type check further down when it is encountered - this would avoid creating a stream for every single field.
Is the name check still necessary ? Wouldn't the type check be enough.
There was a problem hiding this comment.
@costin I have pushed a new commit with a slightly different approach.
* Do not ignore fields whose names start with underscore, unless they are meta fields. * Filter out _size field. (cherry picked from commit 347468e)
This is a fix for #36206.
The solution is to ignore the fields names starting with
_only if their type names also starts with_(a particularity of ES meta fields).