Request for minor tweaks to `0.1.0`

[richb-hanover]

Crusader 0.1.0 is working well. Some fussy requests:

• If I forget to start the server on the other machine, I immediately get an error like this. It would be good to ask, "Is there a server running there?" (I think there are situations where this message does appear...)

  Error: Failed to connect to server
  
  Caused by:
      Operation timed out (os error 60)
  

• If I enter the address of a non-existent device in the GUI, I get the same message, except with long timeout - 75 seconds? A shorter timeout would be good. (Presumably there's some "liveness" test that differentiates between these two situations.)

• In the plots, it would be good to place the colors in the "time order" that is:

  • Download, Upload, Aggregate for throughput
  • Down, Up, Round-trip for latency

• In the plot created by crusader remote web GUI, the time stamp at the upper-left corner should be YYYY-MM-DD hh:mm:ss (the delimiters are currently swapped).

• (Longer feature request, perhaps) As I make continued tests, I have trouble remembering which plot goes with which test condition. How could the plots contain more identifying info: date&time (although they are in the filename), client and server address, a "title" string that replaces "Latency Under Load" at the top, etc.

Thanks as always.

[richb-hanover]

In the Latency Test tab:

• It would be good to allow Return to begin the test (as with the Client tab)
• It would be useful to allow Return (perhaps any key press) to pause the latency test. I saw something interesting that I wanted to memorialize, and lost it fumbling around trying to click the Stop button

[richb-hanover]

These comments are still related to 0.1.0 pre-built binaries on macOS 12.7.5.

• There is a latency test at startup that generates the "Idle latency to server xxx ms" message. Sometimes this is incorrect [1]. I suggest:
  • Test the remote server with three UDP pings, and take the minimum value
  • Display, "Minimum idle latency to server xxx ms"
  • It's a reasonable change to display the minimum time because that sets the floor below which latency would be unlikely to go. Throwing away a couple larger values ignores the human latency time (from, say, Little Snitch) or one aberrant Wi-Fi response.
• (very fussy comment) The baselines of the "Server Address" label and the IP address field of the Client tab don't quite line up.
• The same thing (unaligned baseline) happens after clicking the "Save as Image" button.

[1] I use the Little Snitch network monitor and application firewall. My suspicion is that it jumps up and requires me to authorize a connection to a new network address. That causes Crusader to include the time it takes me to click OK in the measured latency. (Sometimes it takes me seconds to respond, and the measured "idle latency" is roughly that time.)

Thanks as always!

[Zoxc]

• Sometimes this is incorrect [1]

It does sounds like that would be your firewall buffering the packets. Crusader currently uses the median of up to 50 samples as the idle latency.

• (very fussy comment) The baselines of the "Server Address" label and the IP address field of the Client tab don't quite line up.

Yeah, that's an egui bug.

[richb-hanover]

Closing - most have been implemented