Runtime threads, CMake, fixed warnings #6
Conversation
…n well-defined C++ code
Dynamic parameters for schedulers
Explore the appeal to non-valid memory in history
Cmake build system
Fix warnings from clang compiler
Remove test_suite
Backward compatibility has been supported
|
Hi @dorooleg, Thanks for improving things. There are quite a few patches to review at once. I also see some merges, which I prefer to be removed (let's keep history linear). Also I see a commit which conflicts with your recent commit that is already merged, this needs to be resolved. Please split commits into logical smaller groups. What's the motivation for making thread counts dynamic? How does this affect performance? |
.gitignore
Outdated
| @@ -30,3 +30,6 @@ | |||
| *.exe | |||
| *.out | |||
| *.app | |||
|
|
|||
| # QtCreator | |||
There was a problem hiding this comment.
Why? How does QtCreator involved in relacy?
| *.dll | ||
|
|
||
| # Fortran module files | ||
| *.mod |
There was a problem hiding this comment.
Fortran?
Please let's keep only the stuff that actually in use here.
From object files we need only *.o I think. What is *.gch precompiled headers? Who creates them? Do we create any dylib/dll? Do we build any static libraries at all?
relacy/context.hpp
Outdated
| #include "random_scheduler.hpp" | ||
| #include "full_search_scheduler.hpp" | ||
| #include "context_bound_scheduler.hpp" | ||
| #include "schedulers/random_scheduler.hpp" |
There was a problem hiding this comment.
Commits like this need some explanation of the motivation.
Scheedulers structure changed. Ok, good. But why? Can we not do this change? What is the problem this commit solves?
@dvyukov, what do you think is better to create a lot of little PR or use the current big?