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
It looks like fmt 5.1.0 introduced the fp class, which includes a ctor that calls bit_cast<uint64_t>(d). This ctor gets called from various places such that the type of d is double. In fmt 5.1.0 the first place that triggers the static_assert is in basic_writer::write_double. In fmt 5.2.1 the first place is in grisu2_format_positive.
I've been using fmt up to 5.0.0 targeting an architecture that doesn't support 64-bit floats, and so has both float and double defined to be 32-bit. Even though FMT_USE_GRISU is presumably 0 (I didn't define it, and it looks like it defaults to 0), the static_assert still triggers, preventing use of fmt 5.1.0 and 5.2.1 (I haven't tried 5.2.0, but I assume it has the same issue).
The text was updated successfully, but these errors were encountered:
It looks like fmt 5.1.0 introduced the fp class, which includes a ctor that calls bit_cast<uint64_t>(d). This ctor gets called from various places such that the type of d is double. In fmt 5.1.0 the first place that triggers the static_assert is in basic_writer::write_double. In fmt 5.2.1 the first place is in grisu2_format_positive.
I've been using fmt up to 5.0.0 targeting an architecture that doesn't support 64-bit floats, and so has both float and double defined to be 32-bit. Even though FMT_USE_GRISU is presumably 0 (I didn't define it, and it looks like it defaults to 0), the static_assert still triggers, preventing use of fmt 5.1.0 and 5.2.1 (I haven't tried 5.2.0, but I assume it has the same issue).
The text was updated successfully, but these errors were encountered: