Specify output location of bench executable

[sdarwin]

This will build the benchmark executable as ./bin.v2/bench instead of bin.v2/libs/json/bench/clang-linux-14/release/cxxstd-20-iso/link-static/bench. By outputting the exe in the same place each time (similar to how b2 generates it's own file as ./b2) it is easier to find and won't change later.

It would be more convenient if the output locations were ./bin.v2/clang/bench and ./bin.v2/gcc/bench. That requires knowing the 'toolset'. Interestingly from the faqs, "How do I get the current value of feature in Jamfile? This is not possible, since Jamfile does not have "current" value of any feature, be it toolset, build variant or anything else. " $(toolset) doesn't have a value. We can't define the output location to be ./bin.v2/$(toolset)/bench

After merging this into develop branch runbenchmarks.py will break. I will need to modify the script, manually copy bench into ./bin.v2/clang/bench and ./bin.v2/gcc/bench, and run benchmarks from there. @grisumbras if it looks ok, go ahead and merge.

[codecov]

Codecov Report

Merging #946 (e98089b) into develop (66d9251) will not change coverage.
The diff coverage is n/a.

Additional details and impacted files

@@           Coverage Diff            @@
##           develop     #946   +/-   ##
========================================
  Coverage    92.96%   92.96%           
========================================
  Files           85       85           
  Lines         8056     8056           
========================================
  Hits          7489     7489           
  Misses         567      567           

---

Continue to review full report in Codecov by Sentry.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 66d9251...e98089b. Read the comment docs.

[sdarwin]

https://benchmark.cppalliance.org/ processes commits going back to 2020 in order to compare them. This pull request won't be able to retroactively modify earlier copies of the bench Jamfile. So runbenchmarks.py will maintain it's own copy of the file, and write it to disk each time. That's in place now.

[grisumbras]

Your change will break any multiconfiguration build that builds project json/bench, because several targets would try to create the same file.

In #947 I instead added a different target that has to be requested explicitly. The target is bench-local. You can get it built with e.g.

b2 toolset=gcc bench//bench-local

Note that for example

b2 toolset=gcc,clang
b2 toolset=gcc bench//bench-local
bench/bench-local bench/data/*
b2 toolset=clang bench//bench-local
bench/bench-local bench/data/*

will not work as expected, because b2 toolset=gcc,clang builds bench//bench for both gcc and clang, then b2 toolset=gcc bench//bench-local builds bench//bench-local for gcc, and then b2 toolset=clang bench//bench-local checks if bench//bench-local needs to be updated, sees that it's more recent than its dependency (bench//bench) and decides not to update it.

If this is a problem, I can set bench//bench-local to always be considered out of date.

[sdarwin]

Superseded by the pull request "Targets for benchmarks". Closing.