Conversation
|
One question before opening PRs: the change spans two repositories, and thirdparty/naja in kepler-formal points to your fork (nanocoh/naja) rather than najaeda/naja — we're unsure how that fork tracks upstream. How would you like the PRs organized: a PR to |
|
Hello @Emin017 and thank you for your interest in the project! :) As the project is still growing rapidly, we prefer for now that our users will file issues rather than open PRs as it is hard to stabilize across our development. We are trying to be extremely responsive and fast with solving the issues! For issues, we need a clear description of the problem + a reproduction test that can be included in the unit tests after the fix. Will this work for you? |
Got it, I’ll put together a summary report of the issues first :) |
Codecov Report❌ Patch coverage is
📢 Thoughts on this report? Let us know! |
We hit a hard failure when loading SystemVerilog designs containing
initialblocks into kepler-formal. Following the approach used by yosys-slang (evaluate initialization at load time in the frontend, store it asDFF INITmetadata, let downstream consumers just read it), we added initial-value support to SEC: the naja frontend now canonicalizes sized bit-literal parameters to a canonical binary form at the boundary, and SEC extraction harvestsINITintoinitialStateValueByKey, with no engine changes. Covered by unit tests and end-to-end CLI checks.