Cost matrix early abort

[monteirodom]

Shorten down the request delay, when some sources/targets searches are early aborted.

• Decrease remaining_sources_ for all targets when a source search is aborted due to overpassing the threshold
• Decrease remaining_targets_ for all sources when a target search is aborted due to overpassing the threshold

Issue

#3610

[kevinkreiser]

looks like this change is failing a test:

FAIL] matrix
[==========] Running 2 tests from 1 test suite.
[----------] Global test environment set-up.
[----------] 2 tests from Matrix
[ RUN      ] Matrix.test_matrix
/root/project/test/matrix.cc:238: Failure
The difference between results[i].dist and matrix_answers[i].dist is 99996361, which exceeds kThreshold, where
results[i].dist evaluates to 100000000,
matrix_answers[i].dist evaluates to 3639, and
kThreshold evaluates to 1.
result 7's distance is not close enough to expected value for CostMatrix
/root/project/test/matrix.cc:242: Failure
The difference between results[i].time and matrix_answers[i].time is 99996056, which exceeds kThreshold, where
results[i].time evaluates to 100000000,
matrix_answers[i].time evaluates to 3944, and
kThreshold evaluates to 1.
result 7's time is not close enough to expected value for CostMatrix

your description of what you wanted to do made sense to me but it looks like the tests show that now certain pairs of locations that used to get results dont get results anymore (since they come back with that absurdly high number). if you need help learning how to run the tests let me know so you can debug it and get it workign the way you want

[monteirodom]

Yeah, I've seen the tests failing ... no idea why the changes make some good results to be broken.
Was trying to have the tests run locally but got some errors at compilation (on macOS).
Tried on docker, but failed also at 74% of compilation. Here's the last lines of the compilation

#14 2337.1 [ 72%] Linking CXX executable gurka_route_summary
#14 2350.2 [ 72%] Built target gurka_route_summary
#14 2350.3 [ 72%] Generating scripts.log
#14 2350.4 + /usr/bin/python3.8 -m unittest discover -s /src/valhalla/test/scripts -v
#14 2359.9 make[3]: *** [test/scripts/CMakeFiles/run-scripts.dir/build.make:75: test/scripts/scripts.log] Error 1
#14 2360.1 make[2]: *** [CMakeFiles/Makefile2:16848: test/scripts/CMakeFiles/run-scripts.dir/all] Error 2
#14 2360.1 make[2]: *** Waiting for unfinished jobs....
#14 2386.0 [ 72%] Linking CXX executable gurka_route
#14 2392.9 [ 72%] Built target gurka_route
#14 2398.4 [ 72%] Linking CXX executable gurka_route_with_narrative_languages
#14 2400.5 [ 72%] Built target gurka_route_with_narrative_languages
#14 2400.8 [ 73%] Linking CXX executable gurka_search_side_of_street
#14 2414.6 [ 73%] Built target gurka_search_side_of_street
#14 2426.2 [ 73%] Linking CXX executable gurka_simple_restrictions
#14 2427.9 [ 73%] Linking CXX executable gurka_service_turn
#14 2428.0 [ 73%] Built target gurka_simple_restrictions
#14 2429.3 [ 73%] Built target gurka_service_turn
#14 2430.8 [ 73%] Linking CXX executable gurka_phonemes
#14 2432.0 [ 73%] Built target gurka_phonemes
#14 2432.0 [ 74%] Linking CXX executable gurka_stop_signs
#14 2433.3 [ 74%] Built target gurka_stop_signs
#14 2433.4 make[1]: *** [CMakeFiles/Makefile2:2404: test/CMakeFiles/check.dir/rule] Error 2
#14 2433.6 make: *** [Makefile:735: check] Error 2

Took off the gurka and scripts testing parts, then the tests do compile now. On the way to check the errors ...

[kevinkreiser]

this looks good to me, @monteirodom can you put a changelog entry linking this pr to the bottom of the enhancements list. thanks again!

[dnesbitt61]

did a quick comparison using valhalla_run_matrix and it looks good.

[kevinkreiser]

looks like theres a conflict with master, probably in the changelog and i dont have permissions to fix it on your branch. but once you resolve the conflict we can get your PR merged! thanks again for hanging in there!

[kevinkreiser]

can you move this to the end of the list please?

[monteirodom]

Sure, done.