Hello,
If I use ETL to import an edge with a compound unique index, then even if
the index has metadata:{mergeSameKeys:true} and the edge has
skipDuplicates:true, then the duplicate edge will still be created.
If I run two sequential ETL imports with this data :
https://gist.github.com/bmcgavin/69be0b2ebbf535fdf3e9 and this import json
: https://gist.github.com/bmcgavin/177b549a68c7e38e6626 I would expect
there to be one Link and two Articles - effectively the second ETL import
is null because the data is already present.
However I get two Articles and two Links, because the duplicate Edge is not
deleted.
I think the problem is in OIndexUnique.java, line 69-71 :
if (mergeSameKey != null && mergeSameKey)
// IGNORE IT, THE EXISTENT KEY HAS BEEN MERGED
;
But at this point the Edge has already been created, it's just the Index
that will be skipped. I can't confirm this, because I'm not sure how to get
the contents of a compound index. SELECT FROM index:Article.articleId WHERE
key = '12345' works, but I don't know the working version of SELECT FROM
index:Link.in_out WHERE key = '#11:1_#11:2'. However, SELECT FROM Link
gives two Edges with matching in and out values, which should be invalid.
I have generated a version of core with this line :
if (mergeSameKey != null && mergeSameKey)
// Need to delete the newly created edge
iSingleValue.getRecord().delete();
And I now with two ETL runs I get two Articles and one Link, as expected.
I'm not sure if deleting the edge is the right thing to do, or if there's a
way to check the index before creating the Edge?
Thanks,
Rich
--
---
You received this message because you are subscribed to the Google Groups
"OrientDB" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to [email protected].
For more options, visit https://groups.google.com/d/optout.
