Historical changelog¶

The narrative history of the project through 1.0.5. From 2.0.0-rc.1 onward see the generated Changelog.

Revision 2.0.0¶

Renamed the public API from camelCase to snake_case (addSources becomes add_sources, getMibVariants becomes get_mib_variants, and so on). Every former name still works and warns with DeprecationWarning; a subclass that overrides a former name is installed under the new one, so existing overrides keep being called
Generated MIBs are now reproducible: import and export symbol lists were emitted in set iteration order, so the same source produced different output from run to run
The generated header no longer records the build clock, the interpreter version or the absolute source path. It carries Source digest sha256:... instead, taken over the ASN.1 text with newlines normalised, so one MIB has one digest on every platform
MibInfo gained a digest field holding that value
The JSON backend now emits the stand-in column it synthesises for an SMIv1 INDEX naming a bare type, matching the PySNMP backend
PySMI no longer imports pwd, so it can be imported on Windows
The JSON backend no longer nests a default value under two default keys. A default now reads at obj["default"], flat like every other clause, instead of obj["default"]["default"]

Removed all Python 2 era code patterns and bumped minimum Python version to 3.10
Migrated build system from Poetry to UV with the hatchling backend
Replaced flake8/black/isort with ruff and pylint (configured in pyproject.toml)
Replaced sys.exc_info() with except ... as exc and removed try/except ImportError import shims (importlib/imp, json/simplejson, OrderedDict)
Dropped the unittest2 fallback in favor of direct unittest with pytest as the test runner
Updated CLI script shebangs to python3 and refreshed the Copilot instructions to match the new toolchain

Bumped upper Python version to 3.7 and enabled pip cache
Exit code indication of the command-line tools aligned with sysexits.h to report more useful termination status

Fixed pysnmp lower version in test-requirements.txt
Fixed compiler crash when building comments at a platform which has broken users/groups databases

The mibcopy tool implemented to copy MIB modules from files with potentially messed up names into a directory under canonical MIB names picking up the latest MIB revision along the way.
ZIP archive reader implemented to pull ASN.1 MIB files from .zip archives pretty much in the same way as from plain directories
HTTP/S proxy support added (through respecting http_proxy environment variable) by switching from httplib to urllib2 internally
Copyright notice bumped up to year 2018
Project site in the docs changes from SourceForge to snmplabs.com
PRODUCT-RELEASE generation added to the JSON code generator
Added special handling of BITS-like DEFVAL syntax for Integers that occurs in buggy MIBs
Fixed missing REVISIONS generations in MODULE-IDENTITY

Added MIB status, product release and revision description set calls at pysnmp code generator
Changed REVISION field format in JSON representation - it is now a list of dicts each with revision timestamp and description text
MIB REFERENCE fields are only exported if –with-mib-text is on
Sphinx documentation theme changed to Alabaster
Multiple fixes to pysnmp codegen not to produce function calls with more than 255 parameters

Fix to SMI lexer to treat tokens starting from a digit as belonging to a lower-cased class. This fixes sub-OID parsing bug (specifically, 802dot3(10006))
Fix to the mibdump.py local MIB path automatic injection in front of existing –mib-sources

INET-ADDRESS-MIB configured as pre-built at pysnmp codegen
JSON codegen produces “nodetype” element for OBJECT-TYPE
Fix to mibdump.py –destination-directory option
Fix to pysnmp and JSON code generators to properly refer to MIB module defining particular MIB object

The @mib@ magic in reader’s URL template made optional. If it is not present, MIB module name is just appended to URL template
Send User-Agent containing pysmi and Python versions as well as platform name.
Fixed missing STATUS/DISPLAY-HINT/REFERENCE/etc fields generation at pysnmp backend when running in the non-full-text mode
Fixed broken ordereddict dependency on Python 2.6-

Generate REFERENCE and STATUS fields at various SMI objects
Generate DESCRIPTION field followed REVISION field at MODULE-IDENTITY objects
Generate PRODUCT-RELEASE field at AGENT-CAPABILITIES objects
Generated Python source aligned with PEP8
MIB texts cleaned up by default, –keep-texts-layout preserves original formatting
Fix to the ordereddict conditional dependency
Missing test module recovered
Failing tests fixed

JSON code generating backend implemented
Experimental JSON OID->MIB indices generation implemented
Package structure flattened for easier use
Minor refactoring to the test suite
Source code statically analyzed, hardened and PEP8-ized
Files closed explicitly to mute ResourceWarnings
Fixed to Python 2.4 (and aged ply) compatibility
Added a workaround to avoid generating pysnmp TextualConvention classes inheriting from TextualConvention (when MIB defines a TEXTUAL-CONVENTION based on another TEXTUAL-CONVENTION as SYNTAX)