commit.labels and refactoring inconsistencies

[shehan]

In the commit collection there are some records that do not have a refactoring_codebased label even though there is a record in the refactoring collection for that specific commit.

Example:
Query the commit and refactoring collections for the Commit.id = "5e84c0b6770ce37ee9d708b3"
There are 21 records in the refactoring collection
However, in the commit collection, the labels field is empty. Shouldn't the labels field be populated?

A few more commit id's exhibiting this phenomenon:
5e84c208770ce37ee9d771dd
5e84c0f9770ce37ee9d71a89

I am using Smartshark version 2.1 (full db)

[atrautsch]

You are right, the labels are missing and should be there. There was an error with labelSHARK which writes the labels which I missed in the collection phase. Project maven is the only one that is impacted by this.

If you like, you can import the labels separately.

wget http://141.5.100.155/commit_labels_maven.json.gz
gunzip commit_labels_maven.json.gz
mongoimport --db=smartshark_2_1 --collection=commit --mode=merge --upsertFields=_id commit_labels_maven.json