1

I have posted about a similar issue beforehand, where I encountered a compiler error that only popped-up during CI on GitHub using Swift for Linux. Back then, the solution was to replace occurrences of:

guard let x = x else { }

with

guard let x = self.x else { }

In other words: introducing scope was a workaround. But even then, guard let x = x else { } would work most of the time and it only would fail sometimes.

The New Problem

Now I am facing this problem again in this context:

        let (x, y) = myMethod(z)
        guard let x = x else { return self }

which results in the following error:

... error: definition conflicts with previous value
        guard let x = x else { return self }
                  ^
... note: previous definition of 'x' is here
        let (x, y) = myMethod(z)

This works fine in Xcode 13.1 with Swift 5.5.1, but fails again in the GitHub CI workflow where Linux is running. Obviously, in this context I cannot fall back to use scope to tell the compiler which symbol I am talking about and I am forced to use a different name for x in the first line. If I have time, then I might be able to get around to try to replicate this using Docker on my laptop where I could play around with different Linux-versions of the Swift compiler.

The main issue (for me) is not that I have to provide scope and/or rename a symbol, but that the compiler behavior differs across platforms and that it sometimes works fine and fails seemingly randomly.

2

Not sure if it's useful in your context, but you could also do the assignment and the test at the same time:

guard case let (x?, y) = myMethod(z) else { return self }
1 Like
3

Well that is alarming that code which is valid will apparently non-deterministically fail to compile on one platform…

4

I should clarify: if it fails, then it fails deterministically. But you do not know if an assignment will/will not work before actually running the compiler.

Apologies for the confusion.