0.1.6 --- follow openkal 0.13

.github/workflows/ci.yml

@@ -26,7 +26,7 @@ jobs:
 
       - name: Install mcpp
         run: |
-          # ⚠️ A LOOP, BECAUSE ONE `xlings update` CAN RETURN A STALE INDEX
+          # A LOOP, BECAUSE ONE `xlings update` CAN RETURN A STALE INDEX
           # WITHOUT SAYING SO.
           #
           # The index is published as an artifact behind a pointer, and that
@@ -61,7 +61,7 @@ jobs:
       - name: The package builds for a freestanding target
         run: mcpp build --target riscv64-none-elf
 
-      # ⚠️ The load-bearing check, and not "it compiled". These are REPLACEABLE
+      # The load-bearing check, and not "it compiled". These are REPLACEABLE
       # functions: a definition that is present but not exported under the name
       # the compiler emits would compile, link into nothing, and leave every
       # consumer failing on `undefined symbol: operator new` exactly as before.
@@ -75,7 +75,7 @@ jobs:
           echo "exported new/delete symbols: $n"
           test "$n" -ge 12 || { "$NM" --defined-only "$OBJ"; echo "expected at least 12"; exit 1; }
 
-      # ⚠️⚠️ THE TWO DECLARATIONS THIS PACKAGE CARRIES MUST STILL BE openkal's,
+      # THE TWO DECLARATIONS THIS PACKAGE CARRIES MUST STILL BE openkal's,
       # AND UNTIL NOW NOTHING SAID SO.
       #
       # `src/alloc.cpp` declares `kal_alloc` and `kal_free` itself rather than
@@ -86,7 +86,7 @@ jobs:
       # changed either signature, this package would compile, link, and produce
       # a silently wrong calling convention at the only place it matters.
       #
-      # ⭐ SO THE TWO ARE COMPILED TOGETHER, once, in a translation unit that
+      # SO THE TWO ARE COMPILED TOGETHER, once, in a translation unit that
       # ships nowhere: openkal's header and this package's own declarations in
       # the same scope. Disagreeing declarations of the same `extern "C"` name
       # are a compile error that names them, which is the diagnostic wanted.
@@ -95,7 +95,7 @@ jobs:
           set -euo pipefail
           CXX=$(ls -d "$HOME"/.mcpp/registry/data/xpkgs/xim-x-llvm/*/bin/clang++ | head -1)
 
-          # ⚠️⚠️ THE VERSION IS NAMED, AND `head -1' OVER THE STORE IS NOT AN
+          # THE VERSION IS NAMED, AND `head -1' OVER THE STORE IS NOT AN
           # ANSWER TO THIS QUESTION. The package store accumulates every openkal
           # a build on this machine ever needed; the first spelling of this step
           # globbed it and got 0.5.2 -- four minors behind the pin -- and
@@ -138,7 +138,7 @@ jobs:
   # A package that has only ever been built from Linux is a package whose
   # consumers must use Linux, and nothing in these sources says so.
   #
-  # ⚠️ A TOOLCHAIN AXIS IS ABSENT HERE, AND THAT IS MEASURED RATHER THAN
+  # A TOOLCHAIN AXIS IS ABSENT HERE, AND THAT IS MEASURED RATHER THAN
   # ASSUMED. The row for a bare-metal triple names its compiler, and the
   # command-line override does not displace it: `--toolchain gcc@16.1.0` on a
   # `riscv64-none-elf` build resolves llvm@22.1.8 regardless. A matrix over
@@ -147,7 +147,7 @@ jobs:
   # choice is real — openkal, whose declarations are compiled by three families
   # on three systems.
   #
-  # ⚠️ BUILD ONLY, AND DELIBERATELY. Behaviour is asserted once, above, under an
+  # BUILD ONLY, AND DELIBERATELY. Behaviour is asserted once, above, under an
   # emulator. Booting the same image from three systems would be a statement
   # about the emulator rather than about this package, and "the image does what
   # the README says" does not become more true for having been observed from
@@ -188,7 +188,7 @@ jobs:
 
       - name: Install mcpp
         run: |
-          # ⚠️ A LOOP, BECAUSE ONE `xlings update` CAN RETURN A STALE INDEX
+          # A LOOP, BECAUSE ONE `xlings update` CAN RETURN A STALE INDEX
           # WITHOUT SAYING SO.
           #
           # The index is published as an artifact behind a pointer, and that
@@ -222,7 +222,7 @@ jobs:
 
       - name: The allocator cross-builds
         run: |
-          # ⚠️ TWICE, AND THE FIRST IS ALLOWED TO FAIL — every row of this
+          # TWICE, AND THE FIRST IS ALLOWED TO FAIL — every row of this
           # matrix is a machine that has never targeted this triple, which is
           # precisely where mcpp's lazy install of the target C library shows.
           # The first build compiles sources needing its headers before they

README.md

@@ -10,7 +10,7 @@ std-freestanding = { version = "0.3.1", features = ["alloc-kal"] }
 
 That is the whole of the consumer's side.
 
-⚠️ **0.3.1 and not 0.3.0.** The `[feature-deps]` entry in 0.3.0 named a version
+**0.3.1 and not 0.3.0.** The `[feature-deps]` entry in 0.3.0 named a version
 selector the resolver does not have, so the feature activated and its
 implementation could not be fetched. A feature whose implementation cannot be
 fetched is a feature that does not exist. The feature both states the
@@ -50,15 +50,15 @@ naming a mangled operator, which is the same defect with a worse message.
 [openkal][kal] 0.9.0 declares them. On bare metal the board package supplies
 them — the console and the heap region are board facts.
 
-⭐ Those two are all this package reaches for, which is why it follows openkal's
+Those two are all this package reaches for, which is why it follows openkal's
 minor versions without changing: 0.9.0 withdrew `kal_io_result`, made the
 capability words operations and added `kal_memory_granularity`, and none of that
 is on the allocation path. The dependency is declared so that a mismatch is
 reported when the graph resolves, and continuous integration compiles the two
 declarations this package carries against the header they came from, because a
 version dependency catches a version mismatch and not a signature one.
 
-⚠️ The unsized `operator delete` overloads have no size to pass and give zero,
+The unsized `operator delete` overloads have no size to pass and give zero,
 which openkal defines as "not stated". An implementation that needs the size in
 order to free must therefore record it itself; one built over a C library's
 allocator, as the reference bare-metal backend is, does not.

mcpp.toml

@@ -10,7 +10,7 @@
 [package]
 namespace   = "mcpplibs"
 name        = "std-freestanding-alloc-kal"
-version     = "0.1.5"
+version     = "0.1.6"
 description = "The replaceable allocation functions for the freestanding subset, forwarded to openkal"
 license     = "Apache-2.0"
 authors     = ["mcpplibs"]
@@ -21,12 +21,12 @@ provides    = ["freestanding-allocator"]
 # mismatch into a resolution-time message instead of a link-time one, and
 # records what this package's twelve definitions are written against.
 #
-# ⚠️ An openkal IMPLEMENTATION is not declared here and must not be. Which
+# An openkal IMPLEMENTATION is not declared here and must not be. Which
 # implementation is present is a property of the target — on bare metal the
 # board package supplies it, because the console address it writes to is a
 # board fact. A dependency here would bind every consumer to one board.
 [dependencies]
-openkal = "0.12.0"
+openkal = "0.13.0"
 
 [build]
 # The same reasoning openkal's own implementations record. No exception may

src/alloc.cpp

@@ -20,7 +20,7 @@
 // the standard specifies, written in the form a replacement implementation is
 // expected to use.
 //
-// ⚠️ TWELVE, AND THE LAST FOUR ARE THE ONES THAT GET FORGOTTEN.
+// TWELVE, AND THE LAST FOUR ARE THE ONES THAT GET FORGOTTEN.
 //
 // Measured on riscv64-none-elf: defining the eight sized and unsized forms
 // leaves the link failing on `operator delete(void*, unsigned long,
@@ -47,15 +47,15 @@ void  kal_free(void* p, size_type size, size_type align);
 namespace {
 // The alignment an unaligned allocation must satisfy.
 //
-// ⚠️ `__BIGGEST_ALIGNMENT__` and not `alignof(std::max_align_t)`: the latter
+// `__BIGGEST_ALIGNMENT__` and not `alignof(std::max_align_t)`: the latter
 // lives in <cstddef>, which would mean depending on the subset package and
 // closing a cycle. The macro is predefined by both GCC and Clang and is the
 // same quantity. Measured: `alignof(__max_align_t)` does not compile here at
 // all — that identifier comes from <stddef.h>, which a package with no C
 // library include path does not have.
 constexpr size_type kDefaultAlign = __BIGGEST_ALIGNMENT__;
 
-// ⚠️ Size and alignment are forwarded to `kal_free` because openkal's contract
+// Size and alignment are forwarded to `kal_free` because openkal's contract
 // requires them: an implementation may be a bump allocator, a slab, or a
 // wrapper over the C library's, and only the first of those can ignore them.
 // The unsized `operator delete` overloads have no size to pass and give zero,