PROTON-2307 Allow access to connection properties in cpp binding

[pjfawcett]

Allow setting of connection properties via connection_options
Allow reading of properties on incoming connections
Add unit test to container_test.cpp

[pjfawcett]

The AppVeyor "Visual Studio 2015" build failed due to a problem with Microsoft vcpkg. It looks like this has now been fixed
Is there a way to force a re-run of the AppVeyor build or do I need to do a "no-op push"?

[jiridanek]

I don't see option to rerun Appveyor, so the empty commit or empty --amend is probably the only way.

[pjfawcett]

This is my first Qpid Pull Request hence I am unfamiliar with the process and would welcome some advice.

Firstly, what should I do about the failed builds? They seem to be related to errors/problems in the build system(s) rather than to changes I have made, but I might be missing something.

Secondly, and connected, do I need to get this PR to pass all the build checks before it is eligible for review, or is there something else I have to do (apart from being more patient) ?

Thanks

[gemmellr]

I believe some of the builds are a little flakey in CI (and there are also some tests enabled in CI that aren't in the default build because of that). If you are confident the failures aren't from your changes then feel free to just say so. Another approach is to give the build a prod and see if anything changes (e.g git commit --amend --reset-author to redo the commit, followed by a force push).

You don't need to get the tests passing to elicit a review, you just need someone with the requisite knowledge to actually come along (which is not me I'm afraid).

[pjfawcett]

Thanks Robbie.  I'll force another push. I was wary of doing it too repeatedly having seen some recent messages about Travis. CI limits (though later messages suggest that isn't relevant)

[astitcher]

@pjfawcett Firstly many apologies it has taken me so long to review this PR.

I really think this is a useful addition to the API.

I spent a while thinking about whether using std::map<> was a good idea in the context of this API, and I concluded that it makes sense even if it is a rather heavyweight object to be returning. Another alternative might be to just return a proton::map<> which is comparatively lightweight, but not all that convenient without first converting it to a std::map just as you did internally.

I've a few necessary changes below , but otherwise I think this is good essentially as-is.

I also note that session and link objects also have properties (but these maybe not as useful as the connection properties).

If you rebase this change on master I think you can drop the Visual Studio hacks and the tests should pass as appveyor have now fixed their VS2015 image.

[pjfawcett]

@astitcher regarding properties on session and link: I'd be happy to implement similar access to those if it was thought to be useful - probably under a follow-on ticket?

[astitcher]

@astitcher regarding properties on session and link: I'd be happy to implement similar access to those if it was thought to be useful - probably under a follow-on ticket?

That would be great! I'll try to get to it much sooner next time!

[astitcher]

@pjfawcett If you squash rebase these 2 commits then I'll be happy to merge. (I can do it too, but it'll be a little easier for you)

[codecov-io]

Codecov Report

Merging #278 (f5a911d) into master (0ee9ad7) will increase coverage by 4.66%.
The diff coverage is n/a.

@@            Coverage Diff             @@
##           master     #278      +/-   ##
==========================================
+ Coverage   79.56%   84.23%   +4.66%     
==========================================
  Files         343       46     -297     
  Lines       43066     2341   -40725     
==========================================
- Hits        34267     1972   -32295     
+ Misses       8799      369    -8430     

Impacted Files	Coverage Δ
ruby/lib/core/message.rb	57.05% <0.00%> (-31.77%)	⬇️
ruby/lib/util/error_handler.rb	59.37% <0.00%> (-28.13%)	⬇️
ruby/lib/codec/data.rb	76.52% <0.00%> (-19.57%)	⬇️
ruby/lib/codec/mapping.rb	88.17% <0.00%> (-1.08%)	⬇️
python/proton/_transport.py
python/proton/__init__.py
cpp/src/decimal.cpp
cpp/src/proton_bits.cpp
c/src/proactor/epoll_raw_connection.c
cpp/include/proton/null.hpp
... and 287 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 0ee9ad7...55ad0e8. Read the comment docs.

[pjfawcett]

Squashed into single commit.
Ticket PROTON-2330 opened for sessions / links