batzor commented 3 months ago

Description

This PR reverts commit 94a293beb7e0bcd1ec8258ef2ce32188b12ca66d which caused cuda build to fail.

The error message is:

Use --sandbox_debug to see verbose messages from the sandbox and retain the sandbox build root for debugging
bazel-out/k8-dbg/bin/external/perfetto/_virtual_includes/perfetto/third_party/perfetto/perfetto.h(4514): error: expected a "," or ">"
                      2
                      ^

./tachyon/base/ranges/algorithm.h(175): warning #445-D: constant "<unnamed>" is not used in declaring the parameter types of function template "tachyon::base::ranges::in_fun_result<I, F>::operator tachyon::base::ranges::in_fun_result<I2, F2>"
                               std::is_convertible<const F&, F2>{}>>
                                                                  ^

Remark: The warnings can be suppressed with "-diag-suppress <warning-number>"

./tachyon/base/ranges/algorithm.h(183): warning #445-D: constant "<unnamed>" is not used in declaring the parameter types of function template "tachyon::base::ranges::in_fun_result<I, F>::operator tachyon::base::ranges::in_fun_result<I2, F2><I2,F2,<unnamed>>() &&"
                               std::is_convertible<F, F2>{}>>
                                                           ^

./tachyon/math/polynomials/univariate/univariate_polynomial.h(123): warning #815-D: type qualifier on return type is meaningless
    constexpr const size_t NumElements() const {
              ^

bazel-out/k8-dbg/bin/external/pdqsort/_virtual_includes/pdqsort/third_party/pdqsort/include/pdqsort.h(170): warning #68-D: integer conversion resulted in a change of sign
          ip = (ip + cacheline_size - 1) & -cacheline_size;
                                           ^

batzor commented 3 months ago

@ashjeong Unfortunately, I couldn't find out why. I made this PR quickly so that CUDA build works for others. I think we can add it back once we figure it out :)

ashjeong commented 3 months ago

Hmm ok I'll approve it! I think it would be nice to mention perhaps in the pr description that this isn't truly resolved, though, and will need fixing in the future, or perhaps create a GitHub issue 🤔 just for tracking.

kroma-network / tachyon

build: fix cuda build #516

Description