[Libre-soc-bugs] [Bug 1033] Implementation and enhancement of "Test API"

[bugzilla-daemon at libre-soc.org]

https://bugs.libre-soc.org/show_bug.cgi?id=1033

--- Comment #5 from Luke Kenneth Casson Leighton <lkcl at lkcl.net> ---
(In reply to Jacob Lifshay from comment #3)

> for ssh, we need to have a cached ssh connection (one per test process is
> fine imo) 

naah.  setup and teardown override is perfectly fine.
do not make this more complex than it has to be.

if it DOES turn out to be a problem THEN as a SECOND step it
can be "improved". but please do not jump immedately to the
"optimised solution" wasting time especially on a fixed budget
on things that have not been demonstrated as necessary.

please be more brutal and pragmatic as a routine matter of course.


btw ssh under remote connection was solved a long time ago (20 years).

look up telnetbase.py https://bugs.python.org/issue708007
note that it includes a TelnetSSH class.



so we don't try to connect for each test, that ssh connection
> should pass the appropriate options to require non-interactive login (if we
> run with pytest -n 24 do you really want to type your passphrase 24 times?!)

Offically Not Our Problem.  absolutely 100% out of scope.
i repeat. if people have not set up ssh-agent to the target
machine or not put a passphrase (i NEVER do, precisely for
this reason) it is their lookout.


> and to cache the failure to use ssh (so it doesn't try to retry creating a

no.  under no circumstances waste time on solving problems that can be
solved by people not being stupid.

we have limited time and limited budget. getting "something that works
if you follow certain steps" (such as using an ssh key that has
no passphrase) is good enough.

-- 
You are receiving this mail because:
You are on the CC list for the bug.