-
Notifications
You must be signed in to change notification settings - Fork 3
Changes between R2 and R3
This is a new document that is intended to describe the primary changes between R2 and R3.
This document must remain accurate. If you want to add items here, please do not "guess" at changes because what appears as a change may be a bug or oversight, not the way we really want it.
The set functions only work on blocks, strings, and bitset datatypes.
Functions affected: DIFFERENCE, EXCLUDE, INTERSECT, UNION, UNIQUE
These functions work on the BLOCK!, STRING!, and BITSET! datatypes but no other series types. If you find this is a problem, please notify RT.
Note also that these functions should operate somewhat faster than in R2. No speed comparisons have yet been made.
The HASH datatype has been removed and replaced with the MAP datatype.
Functions affected: all datatype action functions related to HASH.
The MAP datatype works like a normal dictionary (associative array) type. Key values will be hashed (as necessary), but data values will not. In addition, key values that do not exist will return NONE and are not considered exceptions. See the map datatype for full details.
INTEGER! is no longer supported for PARSE of STRING value.
I don't think anyone uses it (but maybe I am wrong? Let me know.) It is equivalent to:
intchars: charset "0123456789"
So, that's a better approach anyway. (And we should make some default charsets like that defined from boot.)
The change from 32 bit to 64 bit integer! influences:
Care must be taken with any code that assumes the length of the ISSUE! data returned by to-hex.
- Rebol2
>> to-hex 80 == #00000050
- Rebol3
>> to-hex 80 == #0000000000000050
See R3 Blog article for options and information on the new /size refinement.
There are many other changes. We will post them here.
References to other documents describing differences between R2 and R3.