Project

General

Profile

Actions

Task #3565

open

New Bugzilla field request: Regression-By

Added by Gabor Kelemen 8 months ago. Updated 4 months ago.

Status:
New
Priority:
Normal
Category:
-
Target version:
Team - Q1/2022
Start date:
Due date:
% Done:

0%

Tags:

Description

When dealing with regressions, it turns out to be somewhat difficult for one developer to see all their regressions in a list.

Usually when QA finds and bibisects a regression, they add a comment saying "Adding CC to: Jane Doe" and add the developer to CC. Filtering for such is possible, but unreliable. Sometimes the bibisect is a false positive, but since comments cannot be edited, the "Adding CC to:" keeps the bug in the filter results.
Sometimes this commment is missing, although the bug was properly bibisected to be a regression.
Recently I tried to make lists of all regressions by all our developers and ran into the above issues.

To make searching for "regressions by someone" more reliable I'd like to propose adding a new field to bug pages, named "Regression By" (or something similar). This would be an email type field, that could be added to regression bugs.

Actions #1

Updated by Florian Effenberger 7 months ago

  • Assignee set to Xisco Fauli Tarazona
  • Target version set to Q1/2022

Xisco, something for you to look into?

Actions #2

Updated by Xisco Fauli Tarazona 7 months ago

This has been already discussed in the past. The problem with this approach, which we also want to avoid, is to have a field where we finger point developers, it's kind of rude. That's why we have the 'Adding to: Cc' way, which is more discrete IMHO

Actions #3

Updated by Thorsten Behrens 7 months ago

Xisco Fauli Tarazona wrote in #note-2:

This has been already discussed in the past.

Could you reference this discussion, so we don't need to repeat it here?

The problem with this approach, which we also want to avoid, is to have a field where we finger point developers, it's kind of rude. That's why we have the 'Adding to: Cc' way, which is more discrete IMHO

Understood. Holding back til we've checked the older discussion. But case in point is tdf#146795 ...

Actions #4

Updated by Adolfo Jayme Barrientos 7 months ago

How one uses language matters, so let’s try to entirely remove the person/ego thing out of the equation. Instead of “Regression by”, it could be named “Regressed in” or “Bibisect target”, to refocus people to the commit itself, not the person.

Actions #5

Updated by Xisco Fauli Tarazona 7 months ago

Thorsten Behrens wrote in #note-3:

Xisco Fauli Tarazona wrote in #note-2:

This has been already discussed in the past.

Could you reference this discussion, so we don't need to repeat it here?

Unfortunately I couldn't find it. I think in one of the old QA meetings, minutes are in https://wiki.documentfoundation.org/QA/Meetings but I couldn't find it
Anyway, even if it was decided in the past like that, we can always re-evaluate, of course.
If developers don't mind to be mentioned in a text field, i'm happy to have it, I see the value. Maybe it could be just displayed to the users being part of the 'contributors' group, so just developers and people active in QA can see it...

Actions #6

Updated by Gabor Kelemen 6 months ago

Xisco Fauli Tarazona I like the concept of limiting this fields visibility to the longer term contributors.

Actions #7

Updated by Gabor Kelemen 6 months ago

Adolfo Jayme Barrientos wrote in #note-4:

How one uses language matters, so let’s try to entirely remove the person/ego thing out of the equation. Instead of “Regression by”, it could be named “Regressed in” or “Bibisect target”, to refocus people to the commit itself, not the person.

I think this proposal has some other merit: it would be good to have a separate field (such as “Regressed in”) in the bug description to store the link of the bibisected commit, as this would be more ergonomic/convenient/quick compared to digging it out of 20-25 - sometimes contradicting - comments.

But the main idea is to explicitly associate people (at least the professionals, not the casual fly-by contributors) to their regressions, with the hope they are the right experts to fix/have insight to advise others on fixing such bugs.
I agree that it's probably not productive to slap newcomers with their regressions and the implicit expectation that "it would be nice of them to fix it" - this is not the intention here.

Actions #8

Updated by Xisco Fauli Tarazona 5 months ago

I plan to mention it in the ESC tomorrow. If no one complains, I'll implement it. Will paste the discussion here after the metting

Actions #9

Updated by Xisco Fauli Tarazona 5 months ago

  • Bugzilla Field: Regression-By (Xisco)
    - https://redmine.documentfoundation.org/issues/3565
    - To make searching for "regressions by someone" more reliable
    - current way is to full-text-search in comments: "Adding Cc: to"
    - idea: only a restricted group to set it, to avoid a blame game
    - restriction would be not only for setting but also for reading? (Cloph)
    - yes, if possible to implement
    - what's the benefit? (Stephan)
    - so you can query for your bugs more easily (Xisco)
    - support the idea (Thorsten)
    - quite hard to search currently
    - who to put there? author or committer? (Caolan)
    - up to the person who sets the field? (Thorsten)
    - so far always used the author (Xisco)
    - will this allow removing the result of bad bisects? (Miklos)
    - yes (Xisco)
    - nice improvement (Miklos)
    - would you also re-tag old bugs? (Miklos)
    - up to the QA people
    - plan no mass-tagging (Xisco)
    - looks like a sensible approach (Caolan)
Actions #10

Updated by Xisco Fauli Tarazona 4 months ago

a new field called 'Regression By' is now displayed in Bugzilla. I also submitted https://gerrit.libreoffice.org/c/infra/bugzilla/+/132310 to make it visible only for users in the contributor/admin group.
ideally the field should be an email field and could be autocompleted but that is far beyond my perl skills. maybe something for the new web developer ?

Actions

Also available in: Atom PDF