Add build and tests on Linux, Mac, and Windows to GitHub CI

[ssoelvsten]

This addition is heavily based on my contribution to TPIE. Adding this now also exposes me to (new) compilation issues with Clang (and MSVC). Hence, this also is a follow up on #403 and #317 .

Bug Fixes

• Fix ordering of substitute and reduce in the statistics struct. This explains why said value might be unexpectedly 0.

Clang

src/adiar

Warnings

• algorithms/reduce.h:39:5: Definition of implicit copy assignment operator for 'reduce_arc' is deprecated because it has a user-declared copy constructor.

• */count.*: warning: suggest braces around initialization of subobject

  I do not think, these can be fixed, since the extra braces conflict with the simpler path-based version. The real solution is to switch to the request struct.

• algorithms/substitution.h:33:5: warning: definition of implicit copy assignment operator for 'substitute_arc'

Errors / Test Failures

• request class template
  • Cannot use copy-constructor for an array.
  • It seemed like, a zero-sized array does not take up 0 bytes (on MacOS only?). It essentially treats int[] the way GCC treats std::array<int>. Yet, I cannot reproduce this. Until it shows up again, I will not fix it.

test/adiar

Warnings

• io/test_file.cpp:122:49: warning: lambda capture 'curr_path' is not used
• io/test_levelized_file.cpp:855:95: warning: lambda capture 'curr_path' is not used
• io/test_levelized_file.cpp:1279:45: warning: lambda capture 'tmp_path' is not used
• io/test_levelized_file.cpp:1290:45: warning: lambda capture 'tmp_path' is not used
• io/test_levelized_file.cpp:1301:50: warning: lambda capture 'tmp_path' is not used

Windows

src/adiar

Errors

• The value TRUE, FALSE, or NONE (or more?) are already defined in the preprocessor.

• internal/io/node_writer.h(362,48): error C2672: 'std::max': no matching overloaded function found

• internal/data_types/request.h(212,3): error C2503: 'adiar::internal::request<2,0,1>': base classes cannot contain zero-sized arrays

  • internal/algorithms/quantify.h(94,85): error C2988: unrecognizable template declaration/definition
  • internal/algorithms/quantify.h(94,1): error C4430: missing type specifier - int assumed. Note: C++ does not support default-int

• bdd\apply.cpp(59,7): error C3861: '__builtin_unreachable': identifier not found

• internal/algorithms/reduce.h(279,9) 'adiar::internal::__reduce_cut_add': no overloaded function could convert all the argument types

• bdd\build.cpp(49,33)/ (50,3): missing type specifier - int assumed. Note: C++ does not support default-int

• bdd\build.cpp(49,15)/ (86,12)/ (87,12): error C2146: syntax error: missing ';' before identifier

• Lots of issues with CNL

  • call to object of class type 'cnl::post_operator<Operator,cnl::wide_tag<128,uint32_t,void>,cnl::_impl::duplex_integer<cnl::_impl::duplex_integer<unsigned int,unsigned int>,cnl::_impl::duplex_integer<unsigned int,unsigned int>>,void>': no matching call operator found
    • bdd\apply.cpp(108,63)
    • ? (?,?)
    • bdd\restrict.cpp(65,62):
    • zdd\binop.cpp(143,53)
    • zdd\change.cpp(57,60)
    • zdd\complement.cpp(75,66)
    • zdd\expand.cpp(59,60)
    • zdd\subset.cpp(112,108)
  • 'cnl::_impl::from_rep': no matching overloaded function found
    • bdd/bdd.cpp(51,71): message : see reference to function template instantiation 'adiar::bdd adiar::internal::reduceadiar::bdd_policy(const adiar::__bdd &)' being compiled

Warnings

• statistics.cpp(55,32): warning C4129: 'e': unrecognized character escape sequence
• test_request.cpp(22,9): '==': unsafe mix of type 'const ExpectedType' and type 'const ActualType' in operation

[codecov]

Codecov Report

Patch coverage: 88.660% and project coverage change: +0.005 🎉

Comparison is base (659bae1) 97.165% compared to head (1901521) 97.170%.

Additional details and impacted files

@@              Coverage Diff              @@
##              main      #467       +/-   ##
=============================================
+ Coverage   97.165%   97.170%   +0.005%     
=============================================
  Files           80        80               
  Lines         4480      4488        +8     
=============================================
+ Hits          4353      4361        +8     
  Misses         127       127               

Impacted Files	Coverage Δ
src/adiar/cube.h	100.000% <ø> (ø)
src/adiar/internal/cut.h	100.000% <ø> (ø)
src/adiar/statistics.cpp	86.905% <ø> (ø)
src/adiar/statistics.h	100.000% <ø> (ø)
src/adiar/internal/algorithms/count.h	90.000% <33.333%> (ø)
src/adiar/internal/algorithms/substitution.h	94.318% <60.000%> (ø)
src/adiar/internal/algorithms/intercut.h	98.198% <66.667%> (ø)
src/adiar/internal/algorithms/pred.h	95.506% <75.000%> (ø)
src/adiar/internal/algorithms/quantify.h	97.838% <75.000%> (ø)
src/adiar/bdd/if_then_else.cpp	96.140% <80.000%> (ø)
... and 11 more

Help us with your feedback. Take ten seconds to tell us how you rate us. Have a feature suggestion? Share it here.

☔ View full report in Codecov by Sentry.
📢 Do you have feedback about the report comment? Let us know in this issue.

[github-actions]

Benchmark Report 🟢

origin/ci/check-all-platforms is a regression of 0.35% (compared to origin/main).

Minimum running time (s) for 12-Queens:

origin/main	origin/ci/check-all-platforms
18.12	18.18

Raw Data

Running times (s) for 12-Queens:

origin/main	origin/ci/check-all-platforms
18.26	18.31
18.12	18.26
18.19	18.18
18.12	18.18
18.16	18.23
18.18	18.20
18.16	18.23
18.20	18.31
18.22	18.33
18.17	18.20
18.14	18.26
18.19	18.32
18.14	18.19
18.19	18.28
18.19	18.28
18.23	18.26

[github-actions]

Benchmark Report 🟢

origin/ci/check-all-platforms is an improvement of 0.00% (compared to origin/main).

Minimum running time (s) for 9-Queens:

origin/main	origin/ci/check-all-platforms
0.32	0.32

Raw Data

Running times (s) for 9-Queens:

origin/main	origin/ci/check-all-platforms
0.32	0.35
0.37	0.32
0.35	0.36
0.35	0.35
0.34	0.32
0.35	0.38
0.36	0.32
0.39	0.33
0.37	0.32
0.33	0.37
0.37	0.35
0.32	0.32
0.34	0.33
0.32	0.33
0.43	0.33
0.32	0.32

[github-actions]

Benchmark Report 🟢

origin/ci/check-all-platforms is a regression of 0.61% (compared to origin/main).

Minimum running time (s) for 14-Queens:

origin/main	origin/ci/check-all-platforms
619.29	623.05

Raw Data

Running times (s) for 14-Queens:

origin/main	origin/ci/check-all-platforms
758.42	635.83
626.92	625.39
619.70	624.65
619.29	623.05