Casting negative double to uint does not seem to meet the specification, and works inconsistently across platforms

[antonfirsov]

If I'm not missing anything, the following code should throw System.OverflowException according to the Explicit numeric conversions section of the C# language specification:

double a = -1;
uint b = (uint)a;
Console.WriteLine(b);

Instead, it will print uint.MaxValue on x86.
On Arm64, it seems to return 0 based on my understanding of #45249. I don't have a local repro, but the problem seems to be with the following code on CI:

runtime/src/libraries/System.Net.Http.WinHttpHandler/src/System/Net/Http/WinHttpHandler.cs

Lines 1010 to 1011 in 301056d

	keepaliveinterval = (uint)_tcpKeepAliveInterval.TotalMilliseconds,
	keepalivetime = (uint)_tcpKeepAliveTime.TotalMilliseconds

When we are casting Timeout.InfiniteTimeSpan.TotalMilliseconds == -1.0 to uint the conversion seems to result in 0 on Arm64, while (also incorreclty) returning -1 on x86.

[Dotnet-GitSync-Bot]

I couldn't figure out the best area label to add to this issue. If you have write-permissions please help me learn by adding exactly one area label.

[antonfirsov]

I forgot that default context is unchecked so it's fine, need more sleep.

[EgorBo]

If it's 0 on Arm64 then it's a bug 🤔

[antonfirsov]

According to the C# spec the unchecked case goes as following:

• If the value of the operand is NaN or infinite, the result of the conversion is an unspecified value of the destination type.
• Otherwise, the source operand is rounded towards zero to the nearest integral value. If this integral value is within the range of the destination type then this value is the result of the conversion.
• Otherwise, the result of the conversion is an unspecified value of the destination type.

I'm wondering how do we handle the "unspecified value" cases. Is it our equivalent of C++ UB?