[Lldb-commits] [lldb] [lldb] Assorted improvements to the Pipe class (PR #128719)

[Pavel Labath via lldb-commits]

labath wrote:

It looks like that the windows timeout fix didn't work -- the new test woke ended up waking slightly sooner than expected (198 vs 200ms). At this point I'm guessing that this is due to the use different clocks in the test (`chrono::steady_clock`) and in `WaitForSingleObject` (I have no idea which clock that uses). (The reason this fix worked for the MainLoop class is because it contains a loop which measures the remaining time against the steady clock).

I can see two ways to fix this: relax the test to accept a value slightly less than the expected interval; or add a loop to PipeWindows to wait until the the specified amount of time really elapses. I'm sort of in favor of the second option because it gives a nice guarantee, although I don't expect anybody but tests to care about the few missing milliseconds. Any other opinions?

https://github.com/llvm/llvm-project/pull/128719