Missing letters

[Phil Last]

If I generate all the letters allowed in Dyalog names below ⎕UCS 10000 - I assume there are none above except the underscored alphabet - and filter and manipulate them through 819⌶ I can get the pairs:

      alpha
ABCDEFGHIJKLMNOPQRSTUVWXYZÀÁÂÃÄÅÆÇÈÉÊËÌÍÎÏÐÑÒÓÔÕÖØÙÚÛÜÝ 
abcdefghijklmnopqrstuvwxyzàáâãäåæçèéêëìíîïðñòóôõöøùúûü þ

There are two gaps, the final majuscule "Þ" - thorn - and the penultimate minuscule "ý" - y-acute. Here is the "completed" set.

      omega
ABCDEFGHIJKLMNOPQRSTUVWXYZÀÁÂÃÄÅÆÇÈÉÊËÌÍÎÏÐÑÒÓÔÕÖØÙÚÛÜÝÞ
abcdefghijklmnopqrstuvwxyzàáâãäåæçèéêëìíîïðñòóôõöøùúûüýþ

Assuming there was a reason to include "Ý" and "þ" as valid names was there a reason not to include their counterparts "ý" and "Þ" and is there a reason to continue not to do so?

My reason is that I want/need to be able to switch case of user assigned variable names - to make them case-agnostic - and these two are the troubling exceptions both to use of 819⌶ and my emulator for Dyalog 14.x.

[AndyS|Dyalog]

The rationale for the set of valid characters is somewhat lost in the mists of time, but we believe that it is based on the basic Code page 437 with some of the Icelandic/Danish/Norwegian characters added. The number that could be added would have been limited by what we could do with the classic ⎕av.

We cannot add any more characters to ⎕av, and we need to keep compatibility with classic interpreters. So that list is for now set in stone.

When eventually we persuade everyone to move to Unicode then we would probably want to change to allowing Unicode code point which are valid in indentifiers (plus Delta and Delta_underscore).