[16.0][FIX] auditlog: Fix Invalid Reading of Stale Database Values #3449
Conversation
|
@StefanRijnhart Requesting a review. |
StefanRijnhart
left a comment
StefanRijnhart
left a comment
There was a problem hiding this comment.
Very nice, thank you! Just a nitpick regarding the flushing.
|
Oh, please squash commits into one. |
Due to the introduction of the ThrowAwayCache, in full log mode when attempting to log a write, the reading of the 'new' values effectively reads the 'old' values because prior changes were not flushed to the database yet at that point. This resulted in missing log lines in the created log entry since the old vs. new value comparison did not indicate any changes.
BT-pgaiser
force-pushed
the
16.0.fix-audit-upt-lines
branch
from
e2e588e to
ba392b2
Compare
|
@StefanRijnhart Can we please merge this if it's OK from your side? Thanks. |
|
We cannot merge this yet, see section 6 of https://odoo-community.org/resources/code. |
|
This PR has the |
|
Dear BT members, please consider to not pile up these inter-team reviews as you did here. It feels like you are playing the system. It would be different if the reviews come from known and appreciated contributors to the OCA but I see that for all of you the review on this PR is the only OCA review you did in the whole month (or, ever). |
|
According to the rules... So max 2 reviews can be done by whoever is not PSC or OCA Core Maintainer, so I guess here is the problem that we did approve it 4 times. I want to clarify our position. We don't just simply "approve", we do take the time to properly review the code, and we all are OCA members, although perhaps not with huge activity. Our organization is not "OCA-first", but we do use OCA modules and contribute mainly with migrations, improvements and fixes from time to time, as well as reviews of the contributions of our colleagues. My question is: Is it a problem that we contribute even if our contributions are not monthly or very frequent? So, what is the expectation from the OCA regarding reviewers? We want to do it better and be more aligned with community expectations. |
8 participants
Due to the introduction of the
ThrowAwayCache, in full log mode when attempting to log a write, the reading of the 'new' values effectively reads the 'old' values because prior changes were not flushed to the database yet at that point. This resulted in missing log lines in the created log entry since the old vs. new value comparison did not indicate any changes.It has been confirmed that without the introduced patch for the database flush, the added test fails as expected.