Support Qt 6.4/6.5 Builds

[data-sync-user]

When trying to build using CMake with Qt 6.5.1, the following error is incurred:

• ld: can't link with bundle (MH_BUNDLE) only dylibs (MH_DYLIB) file '~/Qt/6.5.1/macos/./plugins/sqldrivers/libqsqlite.dylib' for architecture arm64

CMAKE_OSX_DEPLOYMENT_TARGET also needs to be bumped to 10.15

┆Issue is synchronized with this Jira Task ┆Reporter: Matt Lichtenstein

[data-sync-user]

➤ Valentina Virlics commented:

Matt Lichtenstein / Naomi Kirby Is there something QA needs to check here? Thank you!

[data-sync-user]

➤ Matt Lichtenstein commented:

Qt 6.5 will likely require a regression test on the entire application, but I don’t think it makes sense to do that testing under this task.

Naomi Kirby would you agree?

[data-sync-user]

➤ Valentina Virlics commented:

We will mark this as Done and label it “qa-verified“ after the regression period for 2.16 is finalized.

[data-sync-user]

➤ Bianca Hidecuti commented:

Hey Matt Lichtenstein,

We were informed by Lesley Norton that the Qt version should be updated to 6.5.1 on MacOS, iOS, and Android, but we noticed that the Qt version for Android and iOS on the latest 2.17 builds is 6.2.4.

For the Linux Jammy 2.17 build the Qt version is also 6.2.4, while on Linux Lunar the Qt version is 6.4.2.

Could you please take a look at this? Thank you in advance!

[data-sync-user]

➤ Paul Oiegas commented:

Hi team, I want to clearly understand what is required from QA for this request and maybe for further work unrelated to this request.

1. Initially, we understood that this request needs Regression testing, but we got confirmation from Matt that it can wait for the RC week and be tested together with the normal Regression run. However, Lesley Norton later told us on Slack that we shouldn't wait for the RC week to test this, as it needs Regression testing ASAP. Does this request actually need specific Regression testing, or normal Feature testing would be enough?

Normally, Regression testing is performed to uncover possible issues that could appear around the areas that were changed. That's why we are running the Regression suite after all the features landed and were initially tested by QA to see that the newly added features didn't break something in the code around them. While Feature testing means testing just the newly landed functionality. If this specific change indeed requires Regression testing sooner, as it could have broad implications, we are happy to do it since we have the available time due to the release date being pushed to late September.

1. While discussions were had around this request, the team got the idea that this separate Regression testing outside of the RC week might not be a one-time request but rather something that we want to do from now on constantly. If this is the case, we'll probably have to make some adjustments/changes for this to happen, as how the team is organized now and the current flow do not allow that. Please let Rebecca Billings or myself know if this is the case, and we’ll further discuss this to see if we can make it happen.

[data-sync-user]

➤ Rebecca Billings commented:

Is there a good reason to run the Regression before AND after the RC? This will add close to two weeks to the testing time and will significantly impact the release schedule. Adrienne Davenport Betty Fleming Santiago Andrigo

[data-sync-user]

➤ Lesley Norton commented:

Paul Oiegas Thank you for the ping.

1.) There has been a lot of back and forth with Qt versions and I’m sorry for the confusion here. As of about an hour ago we have reverted our builds to 6.2.4 on Android, iOS, and Mac so there are no immediate 6.5.1 testing needs for 2.17. All builds, except for Linux, will be on 6.2.4.

{quote}However, Lesley Norton later told us on Slack that we shouldn't wait for the RC week to test this, as it needs Regression testing ASAP. Does this request actually need specific Regression testing, or normal Feature testing would be enough?{quote}

I was first made aware that regression testing was considered a necessity by Valentina in this comment ( https://docs.google.com/document/d/1BhmFMs246MeUS3o4xIYtVqPV097raTk-KtTo-mx4IdU/edit?disco=AAAA2chusVM ). Prior to her comment I had assumed that the bump to 6.5.1 would require very thorough QA testing though since Qt version changes have typically introduced bugs, but this was not mentioned when I reached out to her previously about testing 6.5 builds generated off of this PR ( https://github.com/mozilla-mobile/mozilla-vpn-client/pull/7604 ).

2.) Thank you, I’ll put some time on your calendar for us to chat.

[data-sync-user]

➤ Valentina Virlics commented:

Lesley Norton We should move this to 2.18 (add corresponding fixVersion) and change ticket status, I would suggest Blocked, but your choice.

Will move it back to “In Progress” and remove the 2.17 “fixVersion”, until you decide what to do.

[data-sync-user]

➤ Naomi Kirby commented:

Do we even need sqlite anymore? I think that was only required for the glean.js local storage, which should no longer exist now that we have migrated Glean to Rust.