Automatic skips for tests using sockets about pytest-socket

[jayvdb]

OK, well this plugin claims to be blocking socket, but it isnt blocking socket.getaddrinfo , and possibly a few other bits which could be useful.

This extra socket blocking could be exposed to the user either as

• extra granularity, e.g. --disable-socket=socket,socket.connect,getaddrinfo,getfqdn,sethostname,etc , or
• extra layers, e.g. --disable-name-lookup

[jayvdb]

Turns out that this wont be especially helpful for my use-case unless it also intercepts name resolution, as that fails first, before sockets are opened.

[jayvdb]

For name resolution, it should be relatively easy to wrap/block socket.getaddrinfo by overwriting it with a new function/class which achieves this.

And it looks like https://github.com/disqus/nose-socket-whitelist might do the trick

[miketheman]

That's an interesting approach, would you consider how this might apply as a config argument that makes sense to a user?

[miketheman]

I've thought more about this, and from a testing behavior, I'd prefer my test suite to fail if someone added a usage of a socket that wasn't expected, as then the test suite would pass providing a false sense of security.

I'll leave the issue open and tagged with "help wanted" if anyone wanted to work on this as an opt-in feature.

[jayvdb]

I had envisioned this only as an opt-in.

[github-actions]

This issue is stale because it has been open for 90 days with no activity.

[github-actions]

This issue was closed because it has been inactive for 30 days since being marked as stale.