After. This happened today. 9.4 was deployed over Easter.
Thanks for sharing that. We have a few more questions for you @Will as we try to understand what’s the current status of that branch.
- Is this a branch that is using the “referenced branch” new code strategy?
- Has this branch been analyzed previously on SQ 9.2?
- Has this branch been analyzed previously on SQ 9.3?
- Repeating the question, just in case: Has this branch been analyzed on SQ 9.4?
- If the answer to the 1st and 4th question are “Yes”, does the following query return any results?
select ncri.* from new_code_reference_issues ncri join issues i on i.kee = ncri.issue_key join components c on i.project_uuid = c.uuid join project_branches pb on c.uuid = pb.uuid where c.name = '<YourProjectName>' and pb.kee='<YourBranchKey>'
- Could you run the following query for analysis on the branch and share its results? (it should not contain any confidential information)
select s.version, s.islast, s.period1_mode, s.created_at from snapshots s join components c on c.uuid = s.component_uuid join project_branches pb on c.uuid = pb.uuid where c.name='<YourProjectName>' and pb.kee = '<YourBranchKey>'
Thanks for the help.
This was a bugfix branch created from Develop. The issues could be found from the Measures tab eg. by clicking another link on the homepage.
All branch references are set for Gitflow as follows:
Master is set to use Previous Version
Develop is set to use Reference Branch to Master
All others use the default Reference Branch setting - Reference Branch to Develop
9.4 was upgraded from 9.2.4 over Easter and all master, develop and any branch analysed in the previous 2 weeks was reanalyzed (over 1500 branches across over 500 projects).
However this branch on this project was only created and analysed in the last few days so was only created and run on 9.4. Here is the history of it:
This was part of a fix to a release that was imminent so once found, the developer notes:
Consequently this has moved on and the 6 issues are are now resolved as False Positives. However, I have run these queries anyway…:
Thanks @Will for all of this information. I will try to reproduce your scenario and investigate what’s happening.
I was able to reproduce the behaviour that you reported: updates on issues should update the issue count on new code. Therefore I’ve created this bug to fix that.
As a workaround, retriggering an analysis on the branch should then result in the right issue counts.
Thanks for the update. This is now appearing in increasing numbers of branches. Today one showed 16 new issues on the homepage, but only 8 when clicked on.
The workaround does fix it, but having to analyse branches twice is problematic.
Can you issue a patch for 9.4 to fix this problem?