y-okumura-isp

[AT_CHECK to TORCH_CHECK - PyTorch 1.5 compatible ](https://github.com/daniilidis-group/neural_renderer/pull/86) How about replacing AT_CHECK by TORCH_CHECK if you use pytorch >= 1.5. I succeeded with torch 1.10.

I'm interested in ROS2 realtype behavior, and checking malloc call(see Memory management in [Introduction to Real-time Systems](https://design.ros2.org/articles/realtime_background.html)). I found some malloc calls such as `rosidl_generator_c__String__init` in `rosidl/rosidl_generator_c/src/string_functions.c`. This is called...

@wjwwood Thank you for reply. > rcutils_allocator_t (snip) is the right way to go. I see. > many of our users that are doing real-time replace the logging implementation anyways...

I understand, thank you.

Thank you for comment. Hmm, I rather want to run `inject_on_return` on RHEL/CentOS 😅 BTW, does `inject_on_return` work correctly on MacOS or Windows?

I don't know much about it, so I'm sorry if I'm wrong. > A global .got.plt table? I have heard it in the security context. Is it related with RELRO(RELocation...

I hope we can run `inject_on_return` on Fedora 31 according to [this gist](https://gist.github.com/y-okumura-isp/65dcec2684aea514fe6d724caea9e8e1). This is the summary: ### Summary - By building the execution file as `Full RELRO` and `PIE`,...

I updated PR. As I updated by `commit suggestion` and forgot DCO, DCO CI failed. So I fixed it by `git push -f`.

Has anyone noticed my PR?

Yes, maybe. Though I don't know it is a right way, I could build by the following patch https://github.com/TRI-ML/packnet-sfm/pull/217