tor-browser

The Tor Browser
git clone https://git.dasho.dev/tor-browser.git
Log | Files | Refs | README | LICENSE

regressions.rst (2162B)


      1 How to Mark Regressions
      2 =======================
      3 
      4 Regressions
      5 -----------
      6 
      7 For regression bugs in Mozilla-Central, our policy is to tag the bug as
      8 a regression, identify the commits which caused the regression, then
      9 mark the bugs associated with those commits as causing the regression.
     10 
     11 What is a regression?
     12 ---------------------
     13 
     14 A regression is a bug (in our scheme a ``defect``) introduced by a
     15 `changeset <https://en.wikipedia.org/wiki/Changeset>`__.
     16 
     17 -  Bug 101 *fixes* Bug 100 with Change Set A
     18 -  Bug 102 *reported which describes previously correct behavior now not
     19   happening*
     20 -  Bug 102 *investigated and found to be introduced by Change Set A*
     21 
     22 Marking a Regression Bug
     23 ------------------------
     24 
     25 These things are true about regressions:
     26 
     27 -  **Bug Type** is ``defect``
     28 -  **Keywords** include ``regression``
     29 -  **Status_FirefoxNN** is ``affected`` for each version (in current
     30   nightly, beta, and release) of Firefox in which the bug was found
     31 -  The bug’s description covers previously working behavior which is no
     32   longer working
     33 
     34 Until the change set which caused the regression has been found through
     35 `mozregression <https://mozilla.github.io/mozregression/>`__ or another
     36 bisection tool, the bug should also have the ``regressionwindow-wanted``
     37 keyword.
     38 
     39 Once the change set which caused the regression has been identified,
     40 remove the ``regressionwindow-wanted`` keyword and set the **Regressed
     41 By** field to the id of the bug associated with the change set.
     42 
     43 Setting the **Regressed By** field will update the **Regresses** field
     44 in the other bug.
     45 
     46 Set a needinfo for the author of the regressing patch asking them to fix
     47 or revert the regression.
     48 
     49 Previous Method
     50 ---------------
     51 
     52 Previously we over-loaded the **Blocks** and **Blocked By** fields to
     53 track the regression, setting **Blocks** to the id of the bug associated
     54 with the change set causing the regression, and using the
     55 ``regression``, ``regressionwindow-wanted`` keywords and the status
     56 flags as described above.
     57 
     58 This made it difficult to understand what was a dependency and what was
     59 a regression when looking at dependency trees in Bugzilla.
     60 
     61 FAQs
     62 ----
     63 
     64 *To be written*