Visitar URL original

perf: use c allocator for shared io (#92) · ChainSafe/zapi@a736d8a · GitHub

Repository navigation

Commit a736d8a

authored

perf: use c allocator for shared io (#92)

## Motivation The shared `std.Io.Threaded` uses its allocator for temporary task and future allocations. Using `page_allocator` sends those allocations directly through page mappings instead of allowing ordinary heap reuse. ## Description Use `std.heap.c_allocator` for shared I/O, matching the DSL's existing general-purpose allocator. The zapi module already links libc. This is an allocator-choice optimization; no performance improvement has been benchmarked. Verified with 45 Zig tests, the example addon build, and 136 JS DSL/addon-isolation tests. Changed-file formatting passes; whole-source formatting reports an existing issue in `src/create_callback.zig`, also reproduced at the base revision. AI assistance: Codex prepared this change and ran the checks.

1 parent b56bf2d commit a736d8aCopy full SHA for a736d8a

1 file changed

src/js
- io.zig

`‎src/js/io.zig‎`

-Original file line number
+Diff line change
@@ @@ -1,7 +1,7 @@ @@
 const std = @import("std");
 const lifecycle = @import("lifecycle.zig");
 -const gpa: std.mem.Allocator = std.heap.page_allocator;
 +const gpa: std.mem.Allocator = std.heap.c_allocator;
 const hooks = struct {
     fn init(instance: *std.Io.Threaded, prev_refcount: u32) !void {

Comments

(0)