Archive for June 9, 2023

Friday, June 9, 2023

Whither Stack Overflow Archives?

AMtwo (via Hacker News):

The job that uploads the data dump to was disabled on 28 March, and marked to not be re-enabled without approval of senior leadership. Had it run as scheduled, it would have completed on the first Monday after the first Sunday in June.

I mention the timing, as this change long pre-dated the current moderator strike and related policy changes.


Update (2023-06-13): David Roberts (via Hacker News):

Given past and current turmoils, the fact that [MathOverflow] is not owned by Stack Exchange Inc and the ever-present option of moving to a new home is present in the minds of some people. However, as I mentioned to the other mods in a recent discussion, we are at risk of being in the position of the fictional 1980s UK Prime Minister Jim Hacker, with the power of the nuclear deterrent, but no clear idea of when, if ever, it could ever be actually used in response to an aggressor. Especially when said aggressor may just use “salami tactics“, making things slowly difficult one small step at a time, none of which necessarily deserve responding with the nuclear option.

MO has the option of just leaving SE Inc, taking its data, and setting up elsewhere. But it is our “nuclear option”. Certainly it would not be a good look if MO left SE in the middle of a network-wide dispute of this or that nature (I’m not saying it’s what is under discussion, at present), but it would be a big deal to get everything going again elsewhere.

CloudKit and the iCloud Drive Switch


When I disable iCloud drive [the application] stays, suggesting that access to iCloud is still available.

However, when iCloud drive is disabled, CKContainer.accountStatus returns noAccount, even though I am signed in. When I ignore that and run a query, it returns no records.

Jaanus Kase:

There is one major platform limitation I’ve run into: when you disable iCloud Drive, this also disables iCloud (CloudKit) access to apps like Tact, even though the UI indicates otherwise.

Aaron Pearce:

From the WWDC CloudKit lounge:

As of iOS 17 and macOS Sonoma, disabling the iCloud Drive switch will no longer disable syncing in your app. It’ll be controlled by the individual switch for your app.

This is such a good change.

Greg Pierce:

This is a big deal for sync on corporate devices that often have iCloud Drive disabled.


Feedback Through an Intermediary

Ole Begemann (quoting Accidental Tech Podcast):

Anonymous Apple engineer on Apple’s bug reporting process : “If I get a bug from a developer and want to ask them a question, I can say, please ask the dev a technical question XYZ, and then assign the radar to a black hole. I have no idea if my technical information will be conveyed the way I wrote it. I can’t see what the developer says, other than the initial report. Everything else is through an intermediary.”

What a fucked-up process.

I have a habit of filing bugs early and then adding more information in successive comments as I discover it. But if the Apple engineer only has access to the initial report, perhaps I should delay my reports so that everything can be included together. And there seems to be no point in updating old bug reports to mention that that the issue still reproduces on the latest beta. Maybe these should be refiled as new feedbacks.

Robert Atkins:

If this is accurate the bug reporting process a) doesn’t work for Apple devs, b) doesn’t work for Apple employees. So who does it work for?!

Each year before WWDC, Apple posts a note encouraging developers to file feedbacks. In the past, I used to see Apple engineers emphasizing how important this is and developers chiming in about how to do a good job writing bug reports. This year, sadly, the most common reaction seemed to be laughter. People can’t believe that Apple is acting like we should take this process seriously.

I emphasize “process” because I think most Apple engineers and outside developers do care deeply about fixing bugs but that the system set up by Apple’s leadership to a large extent prevents this from happening.

I still file feedbacks, but probably just a few percent of the ones that I should be writing. I encounter enough issues that I could probably keep myself busy full time reporting them. But my expectation is that almost everything I write will likely be ignored, so I only file bugs that I consider very important and/or that are very easily reproduced with a sample project. I often learn something putting together the sample project, and I try to incorporate it into my unit tests so that I’ll find out if the bug is unexpectedly fixed or the API starts to fail in a different way. Also, if my bug becomes part of a mass-reply where Apple wants me to verify that it still exists, I can do that without spending a lot of time. Unfortunately, many bugs cannot be isolated and easily reproduced in this way.

Alex Rosenberg:

Every project thinks it’s a special snowflake and needs to be specially secured from other projects. This means that even employee bug reports feel like a waste of time that go into a black hole because they can’t be checked up on.

Óscar Morales Vivó:

what you describe in the original post (i.e. you’re “seeing”) is something that happens all the time due to Apple’s culture of (over)secrecy.

I actually think there’s less barriers between outside reports and engineers than there used to be back when I was inside but that doesn’t help if anything secret (rightly or wrongly) is deemed to be tangentially related to an issue.

Alex Rosenberg:

Feedback Assistant is a giant barrier between employees and external people. It solves GDPR problems so that Radar can be ‘free from PII.’


Update (2023-06-13): Tom Lokhorst:

It was great to talk to the engineers working on #ScreenCaptureKit at the lab session. They had some useful tips for how to improve our code.

And I could forward all my feedbacks from the past few months that never reached them through Feedback Assistant.

Update (2023-06-19): Helge Heß:

It is 2023 and Feedback Assistant still can’t pick up your credentials from Keychain. Apple does not want feedback (I know very select individual employees do).


Someone finds a clear and easily reproducible bug in a new Apple framework. They refuse to file a Feedback because they think it won’t be read by Apple, even though a senior compiler engineer at Apple was the one who encouraged them to file it in the first place.

Everything is fine. No total breakdown of trust here, no sir.

Update (2023-08-22): Federico Viticci:

While we’re at it: none of ~20 Feedback items I filed in this beta cycle have been officially addressed (but I was privately told they’re appreciated), and widgets – the core feature of iOS 17 – are still losing their configuration on the Home Screen every few days for me.

I like iOS 17 a lot, and I understand Apple is swamped with the Vision Pro, but it feels like this whole beta > Feedback > beta process needs to be fundamentally rethought.