You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
This issue is resulting out of the following discussion: #3942
First, the documentation for textContainsPrefix is incomplete (more exactly: contradictory) for the case, where the search string contains multiple words / tokens (please have a look at the OP of the discussion for details).
I came up with the following plausible behaviour inferred from the single token case:
"For each token in the query string, at least one token in the text string (read: value of the field which is searched) has to be present, where query token is a prefix of text token"
According to @mad the In-memory implementation org.janusgraph.core.attribute.Text#CONTAINS_PREFIX works as I inferred, but SolrIndex and also LuceneIndex behave differently. Also according to @mad this could be considered a bug. For the LuceneIndex we found the actual behaviour (works just like regular tokenized textContains if query string consists of multiple tokens) and a possible fix to get it working as described above.
If I would be in charge of this, I would first propose to agree on the desired behaviour of textContainsPrefix for the multi token case, where I would actually propose the behaviour above. Then I would consistently implement this across all possible index backends and also very importantly: Add a description of the behavour to the documentation. But, I am not in charge (Spoiler: Probably don't have the time for a PR). So what do you as contributors / maintainers think of how to resolve the issue?
Version: at least 0.6.3
Storage Backend: possibly all?
Mixed Index Backend: at least SolrIndex, LuceneIndex
Expected Behavior: textContainsPrefix should behave consistently and reasonable if query string contains multiple tokens. Further, that behaviour should be documented.
Current Behavior: textContainsPrefix behaves differently across backends.
The text was updated successfully, but these errors were encountered:
This issue is resulting out of the following discussion: #3942
First, the documentation for textContainsPrefix is incomplete (more exactly: contradictory) for the case, where the search string contains multiple words / tokens (please have a look at the OP of the discussion for details).
I came up with the following plausible behaviour inferred from the single token case:
"For each token in the query string, at least one token in the text string (read: value of the field which is searched) has to be present, where query token is a prefix of text token"
According to @mad the In-memory implementation org.janusgraph.core.attribute.Text#CONTAINS_PREFIX works as I inferred, but SolrIndex and also LuceneIndex behave differently. Also according to @mad this could be considered a bug. For the LuceneIndex we found the actual behaviour (works just like regular tokenized textContains if query string consists of multiple tokens) and a possible fix to get it working as described above.
If I would be in charge of this, I would first propose to agree on the desired behaviour of textContainsPrefix for the multi token case, where I would actually propose the behaviour above. Then I would consistently implement this across all possible index backends and also very importantly: Add a description of the behavour to the documentation. But, I am not in charge (Spoiler: Probably don't have the time for a PR). So what do you as contributors / maintainers think of how to resolve the issue?
The text was updated successfully, but these errors were encountered: