If so, that could be noted in the Ramifications, for clarity. Tucker Taft replied as follows: "Yes, that would be reasonable to add to an AARM note."
An alternative modification (alternative to AI22-0048-1/03, that is) would define the one-parameter Day_Of_Week through the new two-parameter Day_Of_Week, in the same way as the two-parameter Local_Image is already defined. And perhaps a renaming as Local_Day_Of_Week could be added, analogously with Local_Image. Tucker Taft replied to this suggestion as follows:
"The issue here is compatibility. The language didn't define the time zone for Day_Of_Week originally; we know that GNAT interpreted it as the local time zone, but we do not know whether other compilers made that same choice.
I agree with your suggestion of defining Local_Day_Of_Week in terms of the two-parameter Day_Of_Week, and then we could have an AARM ramification that the one-parameter Day_Of_Week could be defined in terms of Local_Day_Of_Week in some implementations. This should probably be a separate AI, however."
(I originally made this comment when reviewing this AI for WG 9 approval, but we then agreed to make this a new issue.)
Is the intent of AI22-0048-1/03 that
Day_Of_Week (D) = Day_Of_Week (D, Time_Zones.Local_Time_Offset (D))?
If so, that could be noted in the Ramifications, for clarity. Tucker Taft replied as follows: "Yes, that would be reasonable to add to an AARM note."
An alternative modification (alternative to AI22-0048-1/03, that is) would define the one-parameter Day_Of_Week through the new two-parameter Day_Of_Week, in the same way as the two-parameter Local_Image is already defined. And perhaps a renaming as Local_Day_Of_Week could be added, analogously with Local_Image. Tucker Taft replied to this suggestion as follows:
"The issue here is compatibility. The language didn't define the time zone for Day_Of_Week originally; we know that GNAT interpreted it as the local time zone, but we do not know whether other compilers made that same choice.
I agree with your suggestion of defining Local_Day_Of_Week in terms of the two-parameter Day_Of_Week, and then we could have an AARM ramification that the one-parameter Day_Of_Week could be defined in terms of Local_Day_Of_Week in some implementations. This should probably be a separate AI, however."