NewYour coding agent can read the release notes before it upgrades.Set up the MCP server →
PyPI
PyPI, #3127 most downloaded on PyPI
JCS - JSON Canonicalization
Latest stable
0.3.0
Last release 16 days ago
06 Nov 2026
Ships unpredictably
gaps range from 3 weeks to 4.6 years
Most releases are documented
notes for 3 of 4 stable releases
Nothing withdrawn
no release was ever pulled
5 years old
4 releases · first in 2021
Drag across the ruler to compare, or tap a release to open it.
Breaking: canonicalize raises ValueError for integers that a double cannot represent exactly or whose canonical form reads back as a different integer
canonicalize raises ValueError for integers that a double cannot represent
exactly or whose canonical form reads back as a different integer (e.g. 2**53 + 1, 2**60, or
1152921504606847000) instead of silently rounding them, which mapped different values to
identical output (RFC 8785, section 3.1, #1). Canonical JSON from JavaScript and other double
based implementations can contain such integers; parse it with json.loads(text, parse_int=float)
to get the same bytes as before. Thanks to jmgurney for the report and MoltyCel for the analysiscanonicalize raises ValueError instead of OverflowError for integers beyond
the IEEE 754 double rangecanonicalize(obj, utf8=False) raises UnicodeEncodeError (a ValueError) for
strings with lone surrogates instead of returning them, like the default utf8=True mode already
did (RFC 8785, section 3.2.2.2)canonicalize raises TypeError for dict keys that are not strings instead of an
accidental AttributeError. Converting them like json.dumps would map {1: "a"} and
{"1": "a"} to identical outputLICENSE (Apache-2.0) and
LICENSE.PSF, because jcs/_jcs.py is derived from CPython's json/encoder.pyuv_build backendYour coding agent can read these notes before it upgrades. Set up the MCP server →
- Fixed project subtitle - Moved changelog - Updated dependencies
Created jcs package from original code
jcs package from original codeblackNothing published for this version
One column per quarter.