April 4 2023 nightly hangs forever

[snomos]

To repeat:

• download latest nightly
• quit previous version
• install new version
• quit new version
• open new version
• switch to the All Repositories view

Result: spinning wheel forever, no message whatsoever, see screenshot:

It has now been runing for at least 10 minutes.

Expected: list of available packages, alternatively the new message that it could not connect, with hints of what to do.

[snomos]

Restarting the app did indeed fix the hang (the list of available packages did appear). The bug is thus that the message about connection issues with a suggestion to restart the app did NOT appear.

[dylanhand]

We took a look into this. Just for our records, this bug existed before the April 4 nightly and was not introduced by recent loading spinner changes.

[dylanhand]

Following up on the discussion of UI refresh bug in #48:

I'm still unable to reproduce this issue. I did however find that one of our dependencies had the following warning that it may pose a hang risk:

It's been updated. I'd be surprised if that was the fix, but can't hurt.

I'll report back when a build is available. Currently the CI is broken because this pipeline was also dependent on altool, so an upgrade to notarytool is needed.

[dylanhand]

Builder fixed. The build script has been updated to use notarytool and there is a new nightly build available (download link).

The above mentioned Hang Risk fix did not fix the hanging issue.

Using both the latest stable version of Divvun Manager, and the new nightly build I still get the following:

• Sometimes it loads as expected
• Sometimes it hangs forever
• Occasionally it shows a blank screen with no spinner

This is after repeatedly quitting and re-opening the app. It seems like it could be a race condition.

Here's an example of what I see repeatedly quitting and reopening the app:

In one case I also got this error:

[snomos]

Good that the build is working again. The only thing I can suggest right now is to add more aggressive logging, to possibly identify places where race conditions could occur. Could Sentry logging be of any help?

[dylanhand]

That's what I'll be looking into next. The frustrating part is that the hang never occurs when running from Xcode - it seems to only happen in a production build. So you're right that logging will be the way to go.