Bugfix when optimizer without attributes

[Wikunia]

Bugfix for #201

This converts the optimizer to an optimizer with attributes in all cases. Haven't added a test for this yet as it creates quite some output from Ipopt.

[codecov[bot]]

Codecov Report

Merging #202 into master will decrease coverage by 0.28%. The diff coverage is n/a.

@@            Coverage Diff             @@
##           master     #202      +/-   ##
==========================================
- Coverage   92.33%   92.04%   -0.29%     
==========================================
  Files          22       22              
  Lines        2126     2012     -114     
==========================================
- Hits         1963     1852     -111     
+ Misses        163      160       -3     

Impacted Files	Coverage Δ
src/bb_inits_and_defaults.jl	100.00% <ø> (ø)
src/MOI_wrapper/results.jl	80.76% <0.00%> (-3.11%)	:arrow_down:
src/MOI_wrapper/variables.jl	73.23% <0.00%> (-1.77%)	:arrow_down:
src/MOI_wrapper/constraints.jl	85.71% <0.00%> (-1.25%)	:arrow_down:
src/util.jl	87.76% <0.00%> (-1.20%)	:arrow_down:
src/MOI_wrapper/MOI_wrapper.jl	83.16% <0.00%> (-1.17%)	:arrow_down:
src/printing.jl	78.94% <0.00%> (-0.60%)	:arrow_down:
src/bb_strategies.jl	99.09% <0.00%> (-0.48%)	:arrow_down:
src/bb_integral_or_branch.jl	96.77% <0.00%> (-0.37%)	:arrow_down:
src/BnBTree.jl	93.43% <0.00%> (-0.26%)	:arrow_down:
... and 10 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 1511404...98ec44c. Read the comment docs.

[Wikunia]

Do you mind to check this as well @umbe0 ? @ccoffrin I would like to figure out why the code coverage fails in quite a lot of cases before merging. Do you think it's okay to not have a test for this as the log would be quite huge or can you think of a way to not print it to the stdout?

[umbe0]

Hi @Wikunia, I updated the Juniper package and I tried to execute my optimization with the setting m = Model(with_optimizer(Juniper.Optimizer; nl_solver = with_optimizer(Ipopt.Optimizer), mip_solver = with_optimizer(Cbc.Optimizer))) but the same error keeps coming out.

[Wikunia]

Hi @umbe0 thanks for trying it out. It's not merged and tagged yet. You have to install it via

(@v1.5) pkg> add Juniper#bugfix-optimizer_constructor

Thanks again!

[umbe0]

Ah, sorry, I did not know. Anyway, I followed your indication and, at least in my case, it run correctly! The solver reaches convergence providing all the outputs and no error message. Thank you very much!

[ccoffrin]

@Wikunia, regarding the test. It is possible to capture stdout and stderr so that you will not see the solver's output. For example you can look at how it is done here,

https://github.com/JuliaOpt/Pajarito.jl/blob/master/test/qptest.jl

But I am also open to the idea of not having an explicit test for this case. We might think about adding it to MINLPTests.jl, as it is part of testing the JuMP API.

[Wikunia]

I've added a test case. But have no idea why codecov is failing.

[ccoffrin]

Coverage looks fine, so I would not worry about the codecov error, I would merge and tag.