In the interest of progress, I am going to concede defeat on the problem of
strings. In the interest of progress, I propose to adopt the following
position:

   - Instances of the BitC type String contains a sequence of Unicode code
   points. This is a well-formedness constraint, and it is a requirement that
   all string manipulation functions maintain it.
   - The internal encoding of a String is unspecified, but will be selected
   by the implementation to optimize for space.
   - Indexing operations on Strings operate in no worse than linear time.
   - Indexing operations will be provided on Strings providing
*views*corresponding to ucs1, ucs2, and ucs4 code units.
   - Instances of the BitC type UString likewise contain a sequence of
   Unicode code points.
   - A UString is indexed only on code points.
   - A UString is required to provide constant-time indexing on code points.

Violent objections?


shap
_______________________________________________
bitc-dev mailing list
[email protected]
http://www.coyotos.org/mailman/listinfo/bitc-dev

Reply via email to