Add tests for warnings when passing block to creation of new IO object #1087
Conversation
herwinw
force-pushed
the
io_new_block_warning
branch
2 times, most recently
from
205beee to
7b9bb21
Compare
|
Looks good. It seems this case is missing also for |
|
https://github.com/ruby/spec/blob/master/library/stringio/new_spec.rb
|
herwinw
force-pushed
the
io_new_block_warning
branch
from
7b9bb21 to
ba2ad56
Compare
|
Added |
| @@ -2,5 +2,13 @@ | |||
| require_relative 'shared/new' | |||
|
|
|||
| describe "UNIXServer.new" do | |||
| it_behaves_like :unixserver_new, :new | |||
| platform_is_not :windows do | |||
There was a problem hiding this comment.
Could you move the platform_is_not outside? Because UNIXServer is not even defined on Windows (actually it is in some cases but we are ignoring that so far to keep things simple). Same below.
There was a problem hiding this comment.
I extended this one a little bit to keep things consistent:
- Replace
platform_is_not :windowswithwith_feature :unix_socket, which looks a bit more generic. The current code had both in use. This could resolve the "in some cases on Windows", dependign on the implementation ofwith_feature. - Move these guards to the outer block in all spec files, and remove them from the shared files.
There was a problem hiding this comment.
Great, that's even better, thank you!
Normalize to `with_feature :unix_socket`, which is more descriptive and generic than checking for windows.
Normalize to `with_feature :unix_socket`, which is more descriptive and generic than checking for windows.
Update the spec for StringIO.new to match the style of the other ones, and include the check to see the block is not used.
herwinw
force-pushed
the
io_new_block_warning
branch
from
ba2ad56 to
61167fd
Compare
I think these were all
IOsubclasses, which means I probably missed half of them.I used the syntax
IO::openinstead ofIO.openin the descriptions of the tests, since that is what the warnings say.