MonkeyPatcher should (optionally) raise AttributeError if the target attribute is undefined

[wallrj]

In Twisted's version, you get AttributeError if you attempt to patch a non-existent attribute.
I like that because tests blow up if I mistype the attribute name or if someone else changes the attribute name:

• https://github.com/twisted/twisted/blob/twisted-16.2.0/twisted/python/monkey.py

That's one of the reasons I use patch in Flocker, even on non-global objects.

The testtools version silently ignores these errors:

• https://github.com/testing-cabal/testtools/blob/master/testtools/monkey.py

Does anyone agree that "check attribute exists" should at least be an option?

[jml]

Either an option or a default on a differently named object (e.g.
SafeMonkeyPatcher). Can't change default on current version as that will
break existing code.

On Tue, 5 Jul 2016 at 11:44, Richard Wall notifications@github.com wrote:

In Twisted's version, you get AttributeError if you attempt to patch a
non-existent attribute.
I like that because tests blow up if I mistype the attribute name or if
someone else changes the attribute name:

https://github.com/twisted/twisted/blob/twisted-16.2.0/twisted/python/monkey.py

That's one of the reasons I use patch in Flocker, even on non-global
objects.

The testtools version silently ignores these errors:

https://github.com/testing-cabal/testtools/blob/master/testtools/monkey.py

Does anyone agree that "check attribute exists" should at least be an
option?

—
You are receiving this because you are subscribed to this thread.
Reply to this email directly, view it on GitHub
#237, or mute the
thread
https://github.com/notifications/unsubscribe/AAHq6rialg0kZ8qZ6xR8NyPvxKj2ShYgks5qSjWEgaJpZM4JE_hK
.