Add hexadecimal literals for 'Byte' #1170
Conversation
libraries/common/bytearray.effekt
Outdated
| val unsigned: Int = b.toInt.bitwiseAnd(0xFF.toInt) | ||
|
|
||
| val high = unsigned.bitwiseShr(4).bitwiseAnd(0xFF.toInt) | ||
| val low = unsigned.bitwiseAnd(0x0F.toInt) |
There was a problem hiding this comment.
It would be nice to have bitwise operators on Bytes
There was a problem hiding this comment.
... and equality and stuff :)
jiribenes
force-pushed
the
feat-byte-hex-notation
branch
from
b959d97 to
5f45e95
Compare
jiribenes
force-pushed
the
feat-byte-hex-notation
branch
from
3cf2587 to
29e26d6
Compare
| case Literal((), _) => "()" | ||
| case Literal(n, Type.TInt) => n.toString | ||
| case Literal(n, Type.TChar) => s"'\\${n.toString}'" | ||
| case Literal(b: Byte, Type.TByte) => UByte.unsafeFromByte(b).toHexString |
There was a problem hiding this comment.
This might be a bit confusing, but we can't match on a UByte as it's opaque, so we have to match on a Byte, just to zero-cost convert it to a Byte, just to print the hexString consistently.
... I really don't know how to do this better without either wrapping it in a custom type and dealing with the overhead, or pulling a newtype library :/
There was a problem hiding this comment.
It's ok. You could have toHexString operate directly on Byte and use toUnsignedInt everywhere.
| case Literal((), _) => "()" | ||
| case Literal(n, Type.TInt) => n.toString | ||
| case Literal(n, Type.TChar) => s"'\\${n.toString}'" | ||
| case Literal(b: Byte, Type.TByte) => UByte.unsafeFromByte(b).toHexString |
There was a problem hiding this comment.
It's ok. You could have toHexString operate directly on Byte and use toUnsignedInt everywhere.
| val b = UByte.unsafeFromByte(value) | ||
| // TODO: Literal bytes could now also be just byte-sized in Machine (= use 'UByte'; they are byte-sized in LLVM anyway). | ||
| LiteralByte(unboxed, b.toInt, Coerce(variable, unboxed, k(variable))) |
There was a problem hiding this comment.
I would prefer toUnsignedInt here instead of the newtype wrapping, but ok.
| do emit(0x28) | ||
| repeat(n) { | ||
| do emit(40.toByte) | ||
| do emit(41.toByte) | ||
| do emit(0x28) | ||
| do emit(0x29) | ||
| } | ||
| emitTree(n - 1) | ||
| do emit(41.toByte) | ||
| do emit(0x29) |
| char str[4]; // Byte values range from 0 to 255, 3 characters + null terminator | ||
| snprintf(str, sizeof(str), "%" PRIu8, n); | ||
| char str[5]; // |"0x" ++ <2 hex digits> ++ '\0'| = 5 | ||
| snprintf(str, sizeof str, "0x%02" PRIX8, (uint8_t)n); |
There was a problem hiding this comment.
Shouldn't n already be of type uint8_t? We have typedef uint8_t Byte;. Or is this best practice in C?
Originally by @PhictionalOne, started in #1148 (was on a repo fork).
Resolves #814 by adding hexadecimal literals for bytes:
0xAA,0x00,0xFF,0xA0, etc, exclusively of typeByte.