Issue 652  March 15th 2024

I received two types of emails in response to last week’s introductory comment. The first type had phrases like “No warnings at all? Wow!” and “I’m surprised you didn’t get anything at all!” and the things said in the other type of email were much more along the lines of “You didn’t switch on the checks properly.” 🫣

Yes, it’s true. I made a huge mistake with last week’s intro, where I used enableUpcomingFeature instead of enableExperimentalFeature. It’s crystal clear in the documentation that the former is for Swift 6, and the latter is for Swift 5.9/5.10. I was running with Xcode 15.3, which includes Swift 5.10. 🫢

It’s much more straightforward if you’re switching it on for Xcode, but if you are testing this with a package, there’s a subtle difference between the two.

What tripped me up was when I tried inserting some code I copied from Matt Massicotte’s Concurrency Recipes repository to test that I had the checks enabled correctly. I knew the code would trigger a concurrency warning, and sure enough, a warning appeared as soon as I pasted it in! What I didn’t realise is that that warning would have appeared before I added the strict concurrency flag, too. 😬

I’ve been making lots of silly mistakes recently, and it’s frustrating. I used to have such an eye for detail, and in my mind, I like to think I still do. That said, errors like this are not uncommon for me these days. I don’t think I’m quite old enough yet for it to be age-related, so it’s more likely to be the constant context-switching that I put myself through. I often close a window to find a stack of windows with an embarrassingly high number of half-completed tasks. You’d think I would have got the hang of it by now. 😬

Anyway, the long and short of it is that we do not have zero strict concurrency warnings in the Swift Package Index. We have 526! There appear to be a few common root causes though, so I’m sure that number will come down quickly when we start fixing them. 🛠️

Dave Verwer