Repository navigation
Conversation
cederom
left a comment
There was a problem hiding this comment.
Thank you @anatol please update PR description according to https://github.com/apache/nuttx/blob/master/CONTRIBUTING.md :-)
|
@cederom Thank you for the feedback. PTAL. |
|
@anatol please fix: |
done |
Add a host-side regression program under testing/libc/fmaf that compiles the NuttX fmaf implementation and compares results and required exception flags with host libm across all four rounding modes. Include Make and CMake metadata for building it as a NuttX test app. Keep the comparison harness with the apps testing sources rather than the NuttX tools directory, and document how to invoke it from nuttx-apps. Tests: host C compiler build and run against system libm; 4,072,672 cases passed across four rounding modes. Assisted-by: Claude Code:claude-sonnet-5-5 Signed-off-by: Anatol Pomozov <anatol@google.com>
Thank you @anatol :-)
|
Hello, thank you for your feedback. This test is a split of apache/nuttx#20553 that adds the missing libm function. The code above been used as host testcase for the libm code i added and built/ran it about hundred times at this point. See the commit message part |
There is no mention about new functionality in our Documentation (see [1], [2]). Please add some documentation so people interested in libm / fmaf could read how it works and how to use it. Here is an example PR where author added some code, documentation, and proof of working runtime apache/nuttx#19911 :-)
I could tell you that I ran 1/3 * 4,072,672 cases and you probably would want some logs from me - logs (extracts) really help greatly in the review. Just copy paste most important build and runtime information form the terminal so we know you did build and test the change yourself and that the PR is not fully and only generated by AI with no author verification.. thanks for understanding :-) Why I ask for that? Because your current description states clearly the code was not built and did not run either on host or target board, this may be language issue but also ai hallucination, see:
[1] https://nuttx.apache.org/docs/latest/index.html |
Summary
testing/fmafthat compares NuttXfmaf()with the hostlibm.fmaf()behavior against hostlibm.Impact
fmaf()regression program.lib_fmaf.cimplementation and hostlibm. It is a companion for libs/libm: Add fmaf, using fmadd.s on RISC-V targets with an FPU. nuttx#20553Testing
PR verification Self-Check
☑ This PR introduces only one functional change.
☑ I have updated all required description fields above.
☑ My PR adheres to the Contributing Guidelines and Documentation.
☐ My PR is still work in progress (not ready for review).
☑ My PR is ready for review and can be safely merged into a codebase.
Assisted-by: Claude Code:claude-sonnet-5-5