Phil Frost

Again, want to reiterate I'm just looking for some acknowledgement that these changes would be acceptable. Then I might implement them. I don't want to spend days on a refactor...

Building without cgo breaks name resolution on Linux as well. For example if you have [configured /etc/nsswitch.conf to use systemd-resolved](https://www.freedesktop.org/software/systemd/man/nss-resolve.html#), you'll find without cgo this won't be honored. Instead go...

Does this have any tests asserting that the new functionality works? What's the reason for the special doctest cases in the tests? I also wonder if in some instances an...

On one hand, I can't find anything in [the 2.7 language reference](https://docs.python.org/2.7/reference/simple_stmts.html#raise) that says `raise` must occur within an `except:` block. What it does say is: > If no expressions...

The only difference in Python 3 I noticed was it raises RuntimeError instead of TypeError in the cases where there's no active exception. Otherwise it's the same: the bare `raise`...

As a matter of design principle I can't accept a change that emits an error on legitimate code, even if that code is obscure or odd. Thanks for the PR...

OK, I concede that's a pretty good argument. I'm reopening the PR for consideration. But I also have to get ready for holiday parties so I can't say much more...

Does this need a refresh after those other two PRs were split off?

I don't think pyflakes has a distinction between "warnings" and "errors". It would be helpful to know what version of pyflakes you are using, and more specific info on reproduction.

The willingness of other people to work on the issue is proportional to the effort you put into make a clear, minimal, and easy reproduction scenario. I'm super busy also.