asb added a comment. @ym1813382441 thanks for the contribution. In many cases, this kind of incremental and tightly focused patch is a good way to start things. But in this case, there's a few bigger issues.
- As has been pointed out, the discussion about supporting multiple ISA versions has been started before. The barrier isn't the code change required, but more about agreeing we actually want/need to support multiple extension versions simultaneously (RISC-V International seem to have indicated they'd add new named extensions rather than add instructions to an existing extension in the future), and perhaps the biggest blocker has been around the handling of build attributes, what is accepted / produces an error / produces a warning by what tool and so on. We've been discussing this in the recent RISC-V LLVM sync-up calls and a few people who've been thinking about these issues for a while are hoping to come up with a proposal in the coming weeks. - If the ultimate intention is to support V0.7, then as Craig suggests that's a much bigger discussion. We really need to understand what the level of changes would be required, how this would impact in-tree code for V1.0, what the burden might be going forwards and so on. Not all points are relevant, but I've often thought that the considerations for Contributing Extensions to Clang <https://clang.llvm.org/get_involved.html> provide a good starting point. e.g. - Evidence of a significant user community - A specific need to reside within the upstream tree - A long term support plan - A high quality implementation - Testing Repository: rG LLVM Github Monorepo CHANGES SINCE LAST ACTION https://reviews.llvm.org/D144696/new/ https://reviews.llvm.org/D144696 _______________________________________________ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
