crash on isReadableOrWritable

[shinma]

I'm struggling with one crash since version 0.12.87 (so I'm stuck on 0.12.86 which works fine). It happens almost every time when I start application and in-between running. I was hoping to see fix in each version but it did happen so I'm writing this issue. It always happens when I try to execute isReadableOrWritable. Please take a look at the code below.

Device: any iPhone 64-bit
iOS: 10 or higher
Xcode: 9.2 (9C40b)

    guard !socket.isConnected else {
        throw CommunicationError.connectionError
    }
    
    do {
        try socket.connect(to: hostName, port: port)
    } catch {
        throw CommunicationError.connectionError
    }
    
    do {
        try socket.setReadTimeout(value: 20000)
    } catch {
        throw CommunicationError.connectionError
    }
    
    do {
        try socket.setWriteTimeout(value: 4000)
    } catch {
        throw CommunicationError.connectionError
    }
    
    do {
        try socket.setBlocking(mode: true)
    } catch {
        throw CommunicationError.connectionError
    }
    
    //Write
    let dataSentCount = try socket.write(from: dataToSend)
    
    guard dataSentCount > 0 else {
        throw CommunicationError.socketWriteError
    }
    
    var isReadable = false
    
    //Read
    while !isReadable {
        do {
            let tuple = try socket.isReadableOrWritable(waitForever: false, timeout: 10)
            
            if tuple.readable {
                break
            }
            
            isReadable = tuple.readable
        } catch {
            throw CommunicationError.socketReadError
        }
        
        Thread.sleep(forTimeInterval: 0.5)
    }

I have used Debug version to pinpoint line causing error.

[billabt]

I suspect that this problem and also the problem being seen in issue #125 is related to changes made in PR #110 and it's related PRs. I never saw any of these problems prior to accepting that PR.

@AlanQuatermain: Any ideas what could be causing this? With the rash of bugs that have arisen since accepting this PR, I'm considering rolling it back again.

[AlanQuatermain]

I’m not sure. We have unit tests covering these code paths don’t we? If we can reproduce the issue I ought to be able to resolve everything fairly quickly.

…

Sent from my iPhone

On Feb 22, 2018, at 10:50 AM, Bill Abt ***@***.***> wrote: I suspect that this problem and also the problem being seen in issue #125 is related to changes made in PR #110 and it's related PRs. I never saw any of these problems prior to accepting that PR. @AlanQuatermain: Any ideas what could be causing this? With the rash of bugs that have arisen since accepting this PR, I'm considering rolling it back again. — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub, or mute the thread.

[billabt]

The unit tests are covering most of these code paths but probably not to the extent that it could catch these problems. They seem to edge cases.

Take a look at issue #120. This one is very similar to this case and the user reporting it provided code you can use in a playground. Maybe trying that will give you a clue. I'm still recovering from surgery and unfortunately can't sit (don't ask, 🤪) too long at the computer or I would've tried it before now.

[billabt]

Fixed in 0.12.93.