Removes call to getpeername in SocketStream

[wgwz]

Fixes #609.

I have a few questions:

• What should I do about this test which fails without the getpeername check? https://github.com/wgwz/trio/blob/wgwz/fix-for-609/trio/tests/test_highlevel_socket.py#L27-L30 I'm thinking it might be that the test just doesn't matter anymore.
• What type of a newsfragment should I use? I'm guessing bugfix or removal..

[njsmith]

• Yeah, you can just delete https://github.com/wgwz/trio/blob/wgwz/fix-for-609/trio/tests/test_highlevel_socket.py#L27-L30, it's testing for functionality we've decided to remove
• I'd go with bugfix – the main user-visible effect and motivation for doing this is that it fixes a bug on macOS

[wgwz]

OK cleared up that test and added a newsfragment.

[codecov]

Codecov Report

Merging #744 into master will decrease coverage by 0.01%.
The diff coverage is 100%.

@@            Coverage Diff             @@
##           master     #744      +/-   ##
==========================================
- Coverage   99.34%   99.32%   -0.02%     
==========================================
  Files          96       96              
  Lines       11577    11636      +59     
  Branches      827      832       +5     
==========================================
+ Hits        11501    11558      +57     
- Misses         56       57       +1     
- Partials       20       21       +1

Impacted Files	Coverage Δ
trio/_highlevel_socket.py	100% <ø> (ø)	⬆️
trio/tests/test_highlevel_open_tcp_stream.py	100% <ø> (ø)	⬆️
trio/tests/test_highlevel_socket.py	100% <100%> (ø)	⬆️
trio/tests/test_exports.py	96.42% <0%> (-3.58%)	⬇️
trio/tests/test_ssl.py	100% <0%> (ø)	⬆️
trio/_wait_for_object.py	100% <0%> (ø)	⬆️
trio/tests/test_sync.py	100% <0%> (ø)	⬆️
trio/_sync.py	100% <0%> (ø)	⬆️
trio/_core/_exceptions.py	100% <0%> (ø)	⬆️
trio/socket.py	100% <0%> (ø)	⬆️
... and 5 more

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 106991f...f2261ca. Read the comment docs.

[njsmith]

This is pretty close! A few comments:

• The codecov report is saying there are some new uncovered lines... it looks like it's this getpeername method and this getpeername method, which are little stubs that were only added to make SocketStream happy. Now that SocketStream doesn't care about getpeername, we should delete those methods too. (And I guess update this comment.)

• FYI, you can use Sphinx markup in newsfragments – so instead of ``SocketStream``, it'd be more idiomatic to write :class:`SocketStream`.

• In the contributing guide, we have a bit of discussion about the difference between a good release note and a good commit message. Right now, your newsfragment right now describes what you changed and why, and it would make a great commit message, because commit messages are targeted at other people working on Trio's code. But it's not a great release note, because it's not aimed at the average Trio user. Instead, I'd suggest using something like: "Fixed a race condition on macOS, where Trio's TCP listener would crash if an incoming TCP connection was closed before the listener had a chance to accept it."

[wgwz]

It's nice how the codecov report revealed those issues! I took a look at the reports. Thanks for suggestions on the newsfragments, all points noted. :-)

Oh and sorry it took me so long to finish this.