[FEA]: Improve compile times of `tuple` and `pair`

[miscco]

Is this a duplicate?

• I confirmed there appear to be no duplicate issues for this request and that I agree to the Code of Conduct

Area

libcu++

Is your feature request related to a problem? Please describe.

Compile time analysis of our own CI and external projects like MatX and the rapids libraries shows that tuple and pair are the most expensive types to compile.

While they are relatively simple, there are a lot of SFINAE checks to be done for the various constructors.

We should investigate different techniques, to reduce compile time overhead of tuple and pair

Describe the solution you'd like

Investigate more early returns and using static functions of a struct rather than global inline variables

Describe alternatives you've considered

No response

Additional context

No response

[abdeltaehass]

Hi! I'd like to work on this. I'll first measure the current compile-time cost and study the tuple/pair constructor constraints and existing compile-time benchmarks. Then I'll make a focused change to reduce unnecessary template/SFINAE instantiation while preserving behavior, and provide before/after compile-time measurements and regression tests.

[jrhemstad]

Hi! I'd like to work on this. I'll first measure the current compile-time cost and study the tuple/pair constructor constraints and existing compile-time benchmarks. Then I'll make a focused change to reduce unnecessary template/SFINAE instantiation while preserving behavior, and provide before/after compile-time measurements and regression tests.

Thanks for your interest to contribute @abdeltaehass. However, this requires deep knowledge of CCCL and C++ machinery and is not a good candidate for purely AI-generated contribution without prior context. As such, we are going to keep this issue for core maintainers to work on.