Fix indirect comparisons

[jbcoe]

Do not merge until #478 is submitted.

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 99.55%. Comparing base (2df3748) to head (1772e25).

Additional details and impacted files

@@                         Coverage Diff                         @@
##           dev-jbcoe-no-deferred-requirements     #489   +/-   ##
===================================================================
  Coverage                               99.55%   99.55%           
===================================================================
  Files                                       7        7           
  Lines                                     669      669           
  Branches                                   75       75           
===================================================================
  Hits                                      666      666           
  Misses                                      3        3           

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[nbx8]

All looks good.

[Twon]

I thought we had previously made this change in the LWG session in Tokyo. This is want we want.

[Twon]

Removing the constraints here seems fine to me. Given we return synth-three-way-result I'm not sure what we get by additionally specifying it as convertible to bool.

[jbcoe]

Removing the constraints here seems fine to me. Given we return synth-three-way-result I'm not sure what we get by additionally specifying it as convertible to bool.

Should it become a “mandates” or should we lose it altogether. I’m not sure that there is existing precedent in the library.

[Twon]

Removing the constraints here seems fine to me. Given we return synth-three-way-result I'm not sure what we get by additionally specifying it as convertible to bool.

Should it become a “mandates” or should we lose it altogether. I’m not sure that there is existing precedent in the library.

If we want to be consistent with other types in the standard then looking at the equivalent in std::optional then we would drop it: https://eel.is/c++draft/optional#relops