Iox #33 efficient named pipes with semaphores

[elfenpiff]

Pre-Review Checklist for the PR Author

1. Code follows the coding style of CONTRIBUTING.md
2. Tests follow the best practice for testing
3. Branch follows the naming format (iox-#123-this-is-a-branch)
4. Commits messages are according to this guideline
   • Commit messages have the issue ID (iox-#123 commit text)
   • Commit messages are signed (git commit -s)
   • Commit author matches Eclipse Contributor Agreement (and ECA is signed)
5. Update the PR title
   • Follow the same conventions as for commit messages
   • Link to the relevant issue
6. Relevant issues are linked
7. Add sensible notes for the reviewer
8. All checks have passed (except task-list-completed)
9. Assign PR to reviewer

Notes for Reviewer

Checklist for the PR Reviewer

• Commits are properly organized and messages are according to the guideline
• Code according to our coding style and naming conventions
• Unit tests have been written for new behavior
• Public API changes are documented via doxygen
• Copyright owner are updated in the changed files
• PR title describes the changes

Post-review Checklist for the PR Author

1. All open points are addressed and tracked via issues

References

• Closes TBD

[codecov]

Codecov Report

Merging #847 (efa2a79) into master (e5e91d5) will decrease coverage by 0.17%.
The diff coverage is 66.66%.

@@            Coverage Diff             @@
##           master     #847      +/-   ##
==========================================
- Coverage   75.10%   74.93%   -0.18%     
==========================================
  Files         332      332              
  Lines       11880    11970      +90     
  Branches     2007     2013       +6     
==========================================
+ Hits         8923     8970      +47     
- Misses       2188     2227      +39     
- Partials      769      773       +4     

Flag	Coverage Δ
unittests	73.82% <66.66%> (-0.15%)	⬇️
unittests_timing	30.44% <14.37%> (-0.08%)	⬇️

Flags with carried forward coverage won't be shown. Click here to find out more.

Impacted Files	Coverage Δ
...fs/internal/posix_wrapper/shared_memory_object.hpp	66.66% <ø> (ø)
.../posix_wrapper/shared_memory_object/memory_map.cpp	33.98% <0.00%> (-0.68%)	⬇️
...six_wrapper/shared_memory_object/shared_memory.cpp	68.21% <ø> (ø)
...oryx_posh/include/iceoryx_posh/roudi/roudi_app.hpp	40.00% <ø> (ø)
...ceoryx_posh/source/roudi/application/roudi_app.cpp	72.97% <0.00%> (ø)
..._posh/source/roudi/memory/default_roudi_memory.cpp	90.00% <ø> (ø)
iceoryx_hoofs/source/posix_wrapper/named_pipe.cpp	61.37% <59.82%> (-22.41%)	⬇️
...oofs/source/posix_wrapper/shared_memory_object.cpp	83.33% <92.30%> (+13.69%)	⬆️
...ceoryx_hoofs/internal/posix_wrapper/posix_call.inl	100.00% <100.00%> (ø)
iceoryx_hoofs/source/posix_wrapper/semaphore.cpp	68.99% <100.00%> (+5.08%)	⬆️
... and 7 more

[elBoberido]

alignas(Semaphore) is sufficient.

Wouldn't it make more sense to use two cxx::optional<Semaphore> for this task?

[elfenpiff]

Sadly no since I require direct access to uninitialized memory and the optional does not provide me this kind of access.
The semaphore shouldn't be moved and the idea here is to have the memory available and do a placement new with the creation pattern to avoid the move which would come with the standard create.

[elBoberido]

The list of reason why I don't like this pattern grows. We need to switch to the good old factory method pattern.

[elfenpiff]

@elBoberido but even the factory pattern would be somehow ugly for objects that are neither copyable nor movable.

The user has a place in mind where the object should life and then provides the factory with a pointer to that memory and the arguments.
But the memory handling should be abstracted! Didn't we had the idea for an uninitialized typed array which provides you the raw correctly aligned memory?

This could then be also used as a building block for the list, forward list, vector, optional etc.

[elBoberido]

You are right. This has the same problem. I thought about expected<Error> whatever(optional<Semaphore>& s, ...) could be added to the semaphore. Or maybe something generic where only an emplace method is expected.

The uninitialized array is quite a low building block and shouldn't be overused. After all, you have to take care of calling the dtor by yourself, take care of the alignment and such stuff. We already have a typed container for uninitialized data which takes care of all of that. It's the optional.