Closed CohenArthur closed 3 years ago
I wonder if we should change tests/benchmark.py as well? Since it is a quicker test, I think it matters less.
I think we should keep it as it is now. I like the idea of real-world performance when comparing to httrack
From what I understand we need to launch
local_server_setup.sh
by hand then launchspeed_regression.py
. If this is true, can we try to runlocal_server_setup.sh
fromspeed_regression.py
?
Sure. I wanted to keep it separate so that it's more of a "unix" philosophy. The other issue is that we need to wait for the server to be launched, so it adds a layer of complexity to the script. If we just keep it as this, then two terminals are needed and you can easily see when the webserver is up. We could add something like echo "Web server is up!
in green when launching the web server
I wanted to keep it separate so that it's more of a "unix" philosophy
My idea is to keep them separated but one is calling the other.
If we just keep it as this, then two terminals are needed
That what bothers me :sweat_smile:
The other issue is that we need to wait for the server to be launched, so it adds a layer of complexity to the script
Waiting for "WEBSERVER UP" on stdout seems not that complicated, am I wrong?
That what bothers me 😅
You can also launch it in the background using ./setup_local_server.sh &
if you plan on running tests multiple times. But I would be fine with doing it your way :) I'll tweak it
Waiting for "WEBSERVER UP" on stdout seems not that complicated, am I wrong?
Sure that wouldn't be too complicated haha
You can also launch it in the background using ./setup_local_server.sh &
Yeah, it works. But I sure I will forget it's running in the background haha!
But I would be fine with doing it your way :) I'll tweak it
It's up to you. I can live with it if it's correctly explained somewhere
@Skallwar the speed regression script now launches the webserver locally and handles any exception in order to stop it correctly and not have a zombie webserver
I wonder if we should change
tests/benchmark.py
as well? Since it is a quicker test, I think it matters less.