dev/core#2820: Deduping produces bad results after a rule change

[monishdeb]

Overview

When a dedupe rule is changed, the results appear to be cached.

Reproduction steps

1. Create a dedupe rule on 'Last Name' weight 5, 'First Name' weight 5, threshold 5.
2. Run a dedupe search with this rule - should show contacts with either first or last matching names
3. Edit the dedupe rule, change the threshold to 10.
4. Run a dedupe search with this rule

Before

Shows previous results - those with first or last matching names. Click the 'Threshold' box and note scores of 5 - below the specified threshold of 10.

After

Should only list contacts with matching first and last names where threshold score is 10.

Technical Details

Regression is caused by 9d2f6d5#diff-fe60c89ebe94bbd40114d135459371cc53708db1f2952ade6c0ce9b3468c4dabR2038 this change and affecting since 5.3.0

Fixed by correcting the cache key which is used to fetch and delete the cache of previous dedupe rule.

Comments

ping @eileenmcnaughton @aydun

[civibot]

(Standard links)

• If this is your first pull-request for CiviCRM, please browse CONTRIBUTING.md for information about the development and testing processes.
• If you are reviewing this pull-request, you may wish to consult the test sites and the Review Standards (long template, short template).

[eileenmcnaughton]

thanks @monishdeb this was reported as a regression - should it be against the rc or don't you think it's a regression?

[eileenmcnaughton]

oh I see - it's a very OLD regression - master is fine then!

[monishdeb]

oh I see - it's a very OLD regression - master is fine then!

Yep, master is fine. Also I forgot to mention a workaround, for the previous version, one can click the 'Refresh Duplicates' button on 'Find and Merge Duplicate Contacts' to get the updated list of duplicate contacts.

[aydun]

r-run: PASS

Thanks @monishdeb - that fixes it.

[eileenmcnaughton]

merging based on @aydun r-run