Open alinamoldovan84 opened 3 weeks ago
➤ Norberto Andres Furlan commented:
Issam Mani Razvan Litianu can you take a look to this one? I am worried about this one in particular.
➤ Issam Mani commented:
I was able to reproduce this once and it does seem to be a sync issue since deletion is also propagated to Desktop. Will try to reproduce again and investigate.
➤ Razvan Litianu commented:
Pretty hard to reproduce and follow the steps since it is intermitent but I was assuming it’s retaining cycle. I have a possible fix with the PR ( https://github.com/mozilla-mobile/firefox-ios/pull/22142 )
➤ Issam Mani commented:
{quote}I have a possible fix with the PR ( https://github.com/mozilla-mobile/firefox-ios/pull/22142 ) {quote}
Thanks Razvan Litianu. I had a suspicion that somehow the logged in state for sync is retained somehow because deletion is propagated to desktop, but I could only reproduce this once 🤷♂️
➤ Norberto Andres Furlan commented:
Is this one fixed? Issam Mani Razvan Litianu
➤ Issam Mani commented:
Norberto Andres Furlan Razvan Litianu opened a possible fix. That’s our best bet right now once it’s merged. I will keep trying to reproduce this locally but so far I wasn’t successful.
➤ Nishant Bhasin commented:
https://github.com/mozilla-mobile/firefox-ios/pull/22142 ( https://github.com/mozilla-mobile/firefox-ios/pull/22142|smart-link ) - This PR got merged
Issam Mani lets see how main does for this
➤ Alina Moldovan commented:
This issue still reproduces intermittently using v131 (45674).
➤ Issam Mani commented:
Update:
Made a bit of progress. It seems that we land in a broken state where we signed out of sync/fxa but the code still thinks it’s signed in. Notice the pref says true. This is the only case where the issue happens. I will investigate more to see what we do when we disconnect from sync.
!Group 1.png|width=1792,height=1039,alt="Group 1.png"!
➤ Issam Mani commented:
Ok so update on this. Lougenia Bailey and I investigated this further yesterday ( slack thread ( https://mozilla.slack.com/archives/C0559DDDPQF/p1727363793404829 ) ) and two important findings came up:
The outcome of the discussion is that this is not a big problem to block release of address autofill, since it’s not address specific. It happens rarely and will only propagate the deletion of the addresses/bookmarks deleted when it happens not all of them. Also this is a pre-existing issue as we were able to reproduce this for bookmarks. We can fix this later.
Norberto Andres Furlan Nishant Bhasin what do you think ? This is now the only blocker for address autofill.
Steps to reproduce
Expected behavior
All addresses are synced back on the device
Actual behavior
Sometimes all addresses are accidentally deleted from the user account. This issue is intermittent it happened 4 times out of 6
Device & build information
Notes
Attachments:
https://github.com/user-attachments/assets/33c0187a-f073-4a52-b760-0476b771a07b
┆Issue is synchronized with this Jira Task