You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Right now fiat_backend and simd_backend are mutually exclusive: activating both crate features generates a compile error.
In #414 we discussed various alternatives including something like --cfg dalek_backend="simd". However, since the backends map to optional crate dependencies, features seem like a more natural fit.
As an alternative to a compile error, another option would be to handle the case of both features being activated by choosing a "winner" backend which is selected when both options are enabled.
This isn't a blocker for a v4.0 release though, as it can easily be added afterward as it changes a case which is currently a compile error into one that works.
The text was updated successfully, but these errors were encountered:
Right now
fiat_backend
andsimd_backend
are mutually exclusive: activating both crate features generates a compile error.In #414 we discussed various alternatives including something like
--cfg dalek_backend="simd"
. However, since the backends map to optional crate dependencies, features seem like a more natural fit.As an alternative to a compile error, another option would be to handle the case of both features being activated by choosing a "winner" backend which is selected when both options are enabled.
In #414 I suggested that given various tradeoffs, it should probably be
fiat_backend
because it would be bad to enable a non-verified backend when someone has asked for a formally verified one (even thoughsimd_backend
is superior in pretty much all other respects).This isn't a blocker for a v4.0 release though, as it can easily be added afterward as it changes a case which is currently a compile error into one that works.
The text was updated successfully, but these errors were encountered: