You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Commands take an extremely long time in windows. This is related to the MacOS disparity (See: #1951 ) but at a scale of 10x the disparity.
For example, #1951 illustrates this disparity with bolt --version taking 0.5 seconds on Linux and 1.7 seconds on MacOS. The same command takes 12.5 seconds on Windows.
Similarly, #1951 notes bolt task show taking 1.7 seconds on Linux and 3.3 seconds on MacOS. The same command takes 38.5 seconds on windows.
I routinely run commands that take almost a whole minute.
Expected Behavior
Commands should run within one standard deviation of the average command response time of the environment one is working in. This should apply to Windows, but also MacOS and Linux. I don't think there's as much value in comparing response times in different OSes as there is comparing times across contexts in the same OS. Bolt should respond within the expected response time of a Windows user when in Windows, and the expected response time of a MacOS user on a Mac.
Despite that argument, having a 30-45 second response time for a simple command is an obvious failure of usability.
@johnmetta Thank you for filing this issue. We're aware of performance differences on Windows and have been working on finding ways to improve this. The most relevant open PR you can follow is #2291
Describe the Bug
Commands take an extremely long time in windows. This is related to the MacOS disparity (See: #1951 ) but at a scale of 10x the disparity.
For example, #1951 illustrates this disparity with
bolt --version
taking 0.5 seconds on Linux and 1.7 seconds on MacOS. The same command takes 12.5 seconds on Windows.Similarly, #1951 notes
bolt task show
taking 1.7 seconds on Linux and 3.3 seconds on MacOS. The same command takes 38.5 seconds on windows.I routinely run commands that take almost a whole minute.
Expected Behavior
Commands should run within one standard deviation of the average command response time of the environment one is working in. This should apply to Windows, but also MacOS and Linux. I don't think there's as much value in comparing response times in different OSes as there is comparing times across contexts in the same OS. Bolt should respond within the expected response time of a Windows user when in Windows, and the expected response time of a MacOS user on a Mac.
Despite that argument, having a 30-45 second response time for a simple command is an obvious failure of usability.
Steps to Reproduce
Steps to reproduce the behavior:
Measure-Command { bolt --version }
Measure-Command { bolt task show }
Environment
Additional Context
Filed as a different ticket than #1951 because that seemed to indicate the use of Bundler and a MacOS-specific path of resolution.
The text was updated successfully, but these errors were encountered: