forked from pool/python-pysnmp
- Update to 4.1.14a
- Fix to maxSizeResponseScopedPDU calculation at USM security module: now more precise and robust against screwed values on input. - Fix to MIB loading logic that used to load same-name modules at disticts search paths on each loadModules() call. - Fix to AsynsockDispatcher.runDispatcher() to make use of optional non-default select() timeout value. - AbstractTransportDispatcher now allows user application registering multiple timer callbacks each with dedicated call period. - Asynsock mainloop default idle period reduced to 0.5 sec for better timer resolution. - Fix to SNMPv1->SNMPv2c error status handling at proxy module. This defect may have caused an infinite loop on a multiple var-bind SNMPv1 GetNext operation. - Fix to contextName processing at config.addV1System -- typo rendered passed contextName not committed into LCD. - Fix to unknown ContextName exception handling at CommandResponder App. - config.addVacmUser() now accepts an optional contextName what makes it usable for configuring multiple contextName-bound bases of Managed Objects to SnmpEngine. - MP pending states cache management re-worked so that SNMP engine will now handle an unlimited number of pending request/responses. - Fix to SNMP discovery procedure: include ContentName in SNMP discovery messaging. - Many fixes to AES crypto code that makes it actually working. - Fix to SNMPv2-SMI createUndo operations. - Fix to INFORM sending error handling at oneliner. - Fix to mismatched response PDU handling at CommandGenerator application. - Debug category 'app' (for Application) added to facilitate OBS-URL: https://build.opensuse.org/package/show/devel:languages:python/python-pysnmp?expand=0&rev=10
This commit is contained in:
@@ -1,3 +1,84 @@
|
||||
-------------------------------------------------------------------
|
||||
Tue Sep 14 22:17:17 UTC 2010 - jfunk@funktronics.ca
|
||||
|
||||
- Update to 4.1.14a
|
||||
- Fix to maxSizeResponseScopedPDU calculation at USM security module: now
|
||||
more precise and robust against screwed values on input.
|
||||
- Fix to MIB loading logic that used to load same-name modules at
|
||||
disticts search paths on each loadModules() call.
|
||||
- Fix to AsynsockDispatcher.runDispatcher() to make use of optional
|
||||
non-default select() timeout value.
|
||||
- AbstractTransportDispatcher now allows user application registering
|
||||
multiple timer callbacks each with dedicated call period.
|
||||
- Asynsock mainloop default idle period reduced to 0.5 sec for better
|
||||
timer resolution.
|
||||
- Fix to SNMPv1->SNMPv2c error status handling at proxy module. This
|
||||
defect may have caused an infinite loop on a multiple var-bind
|
||||
SNMPv1 GetNext operation.
|
||||
- Fix to contextName processing at config.addV1System -- typo rendered
|
||||
passed contextName not committed into LCD.
|
||||
- Fix to unknown ContextName exception handling at CommandResponder App.
|
||||
- config.addVacmUser() now accepts an optional contextName what makes
|
||||
it usable for configuring multiple contextName-bound bases of Managed
|
||||
Objects to SnmpEngine.
|
||||
- MP pending states cache management re-worked so that SNMP engine will
|
||||
now handle an unlimited number of pending request/responses.
|
||||
- Fix to SNMP discovery procedure: include ContentName in SNMP discovery
|
||||
messaging.
|
||||
- Many fixes to AES crypto code that makes it actually working.
|
||||
- Fix to SNMPv2-SMI createUndo operations.
|
||||
- Fix to INFORM sending error handling at oneliner.
|
||||
- Fix to mismatched response PDU handling at CommandGenerator application.
|
||||
- Debug category 'app' (for Application) added to facilitate
|
||||
Standard SNMP Applications debugging.
|
||||
- The retryCount semantic of CommandGenerator application changed to include
|
||||
sole retries and do not include initial request. Thus, retryCount=1 will
|
||||
now send up to two packets, not just one as it used to be.
|
||||
- Debugging printout now escapes non-printable characters.
|
||||
- Revision 4.1.13a
|
||||
- UDP over IPv6 transport implemented.
|
||||
- Fix to MIB tree walking code that used to fail on table columns where
|
||||
indices have identical leading parts.
|
||||
- SNMPv1/v2c snmpCommunityTransportTag-based imcoming message filtering
|
||||
implemented (rfc2576).
|
||||
- Revision 4.1.12a
|
||||
- API versioning retired (pysnmp.v4 -> pysnmp).
|
||||
- MIB loading mechanics re-designed to allow ZIP import.
|
||||
- MIB loader supports code objects (py[co])
|
||||
- Installer now uses setuptools for package management whenever available.
|
||||
- The libsmi2pysnmp tool improved to build constraints of more than
|
||||
256 items (Python has a limit on the number of function params).
|
||||
- Missing SNMPTrap PDU API implemented at proto.api.v2c, RFC2576 proxy
|
||||
code reworked.
|
||||
- Fix to sysUpTime OID at SNMPv2 TRAP PDU.
|
||||
Revision 4.1.11a
|
||||
- Twisted integration implemented.
|
||||
- Attempt to use hashlib whenever available.
|
||||
- Fix to oneliner Manager code on < Python 2.4.
|
||||
- Let NotificationReceiver and CommandResponder Apps browsing request details
|
||||
(transport information at the moment) at SNMP engine.
|
||||
- Fix to config.addV1System() to allow multiple systems to co-exist in LCD.
|
||||
- Fix to wrongly succeeding user-parameters-by-community-name searching code
|
||||
in rfc2576 processIncomingMsg() method.
|
||||
- Do sanity checking on PYSNMP_MODULE_ID, Groups and Notifications in
|
||||
libsmi2pysnmp (SF bug #2122489).
|
||||
- Fix to oneliner Notification Originator that sometimes used to send multiple
|
||||
requests at once.
|
||||
- Oneliners LCD names generation code reworked to avoid accidental clashes.
|
||||
- Fix and re-work of sysUpTime value management in LCD.
|
||||
- Fix to pending inform request data caching in mpmod/rfc2576.py -- previous
|
||||
code led to cache data corruption on multple outstanding requests.
|
||||
- In SMI configuration wrapper functions, catch access to non-configured
|
||||
entries and raise an exception.
|
||||
- Allow multuple callback timer functions in all transport dispatchers.
|
||||
- Fix to libsmi2pysnmp code to preserve more underscored object names and
|
||||
to guess the right type between indistinguishable ObjectGroup &
|
||||
NotificationGroup
|
||||
- Fix to MibScalarInstance value setting logic - previous code failed
|
||||
when modifying the same OID multiple times within a single SET operation.
|
||||
- Minor usability improvements to tools/build-pysnmp-mib.
|
||||
- Made MIB objects unexport feature operational.
|
||||
|
||||
---------------------------------------------------
|
||||
Tue Mar 23 13:30:13 CET 2010 - boris@steki.net
|
||||
|
||||
@@ -10,3 +91,173 @@ Tue Mar 23 13:13:13 CET 2010 - boris@steki.net
|
||||
- Added required libsmi require statement and its minimum version
|
||||
for utils sub package main package.
|
||||
|
||||
-------------------------------------------------------------------
|
||||
Mon Feb 16 2009 James Oakley <jfunk@funktronics.ca> - 4.1.10-1
|
||||
|
||||
- Update to 4.1.10a
|
||||
- Internal MIB indexing method __indexMib() unmangled to facilitate
|
||||
setting up mutex there for sharing MIB stuff between threads.
|
||||
- Fixed broken IpAddress value handling in SET operation.
|
||||
- Broken MibBuilder.unloadModules() method now works.
|
||||
- Use getLabel() SMI object method when building MIB tree (in builder.py)
|
||||
in addition to direct attribute access for clearer object protocol.
|
||||
- The MIB building tools updated to match significantly improved
|
||||
smidump tool (libsmi version > 0.4.5).
|
||||
- Made libsmi2pysnmp tool optionally building MIB text fields into pysnmp
|
||||
MIB code (enabled by default) and MibBuilder conditionally loading them
|
||||
up (disabled by default).
|
||||
- SnmpEngine and MsgAndPduDispatcher constructors now optionally
|
||||
take msgAndPduDspr and mibInstrumController class instances
|
||||
respectively to facilitate these objects sharing within a process.
|
||||
- Unique integers, for various parts of the system, are now generated
|
||||
by a nextid.py module. This fixes possible issues with duplicate
|
||||
request IDs and handlers.
|
||||
- Built-in MIBs re-generated to include text fields.
|
||||
- 4.1.9a
|
||||
- UNSTABLE ALPHA RELEASE.
|
||||
- At onliner CommandGenerator, close transport on destruction to
|
||||
prevent socket leak. Implicit async transports registration at
|
||||
default asyncore's socket map has been disabled to avoid side
|
||||
effects.
|
||||
- Fix to rfc2576.v1ToV2c() PDU converter to perform noSuchName error code
|
||||
translation.
|
||||
- Fixes to Notification PDU convertion code at rfc2576 in part of
|
||||
snmpTrapOID handling.
|
||||
- Fix to nonRepeaters object use as sequence slicer (must be int) at
|
||||
cmdrsp.CommandResponderApplication
|
||||
- Make AsynsockDispatcher using its own socket map by default for
|
||||
threading safety. This will break asyncore apps that rely on pysnmp
|
||||
sharing the same socket map with them. A solution would be to either
|
||||
set asyncore map to pysnmp (AsynsockDispatcher.setSocketMap()) or pass
|
||||
pysnmp map (AsynsockDispatcher.getSocketMap()) to asyncore.
|
||||
- Fix to response timeout roundup bug at CommandGenerator and
|
||||
NotificationOriginator code.
|
||||
- Oneline configuration classes made hashable to prevent memory leaks
|
||||
when committing them into CommandGenerator/NotificationOriginator
|
||||
internal repository.
|
||||
- Security information is now released properly in all MP modules.
|
||||
This might fix a significant memory leak.
|
||||
- Typo fix to rfc3411 confirmed class PDU members.
|
||||
- 4.1.8a
|
||||
- UNSTABLE ALPHA RELEASE.
|
||||
- SMI/dispatcher timeout convertion multiplier is actually 100 (1/100 sec)
|
||||
rather than 1/1000. This fix affects timeouts specified through SMI.
|
||||
- __repr__() implemented for UdpTransportTarget, CommunityData, UsmUserData
|
||||
in oneliner module.
|
||||
- Automatically initialize table index values on table management
|
||||
operations (SF bug ID #1671989).
|
||||
- Fix to carrier code: ignore BADFD socket error as it may happen upon
|
||||
FD closure on n-1 select() event.
|
||||
- Fix to MibBuilder.setMibPath() to preserve previously loaded modules
|
||||
intact. Otherwise loadModules() called after setMibPath() might fail
|
||||
with 'MIB file not found in search path' exception.
|
||||
- Fix to oneliner classes that now invoke unconfiguration methods on
|
||||
destruction. This might have caused memory leaks.
|
||||
- Automatically initialize SNMP-USER-BASED-SM-MIB::UsmUserSecurityName
|
||||
columnar object instance on creation, as stated in DESCRIPTION (SF
|
||||
tracker ID #1620392).
|
||||
- Fix to USM timeframe arithmetics (SF bug #1649032).
|
||||
- VACM shortcuts merged into universal add/delVacmUser() to let notifications
|
||||
and requests to co-exist for the same user.
|
||||
- At oneliners, build LCD tables keys from a hashed mix of input parameters
|
||||
to make sure these automatic entries won't interfere or exceed constraints
|
||||
on keys values.
|
||||
- Made use of notificationType parameter of the sendNotification method
|
||||
in NotificationGenerator applications. This parameter used to be
|
||||
ignored in the past. Note, that previously used (and ignored) syntax
|
||||
has been changed in an incompatible way.
|
||||
- Allow plain Python values in setCmd() and sendNotification()
|
||||
methods in CommandGenerator and NotificationGenerator applications
|
||||
respectively.
|
||||
- Multi-target oneliner API-based example script added.
|
||||
- Ignore more socket errors in datagram-type async socket code.
|
||||
- AES cipher now supported (rfc3826).
|
||||
- Fix to messed up tagIDs of noSuchInstance and noSuchObject types.
|
||||
- SET Command Responder fixed to obscure access to non-existing variables
|
||||
by returning notWritable error (SF bug #1764839).
|
||||
- AsynsockDispatcher.setSocketMap() method added to facilitate pysnmp
|
||||
transport integration into third-party asyncore-based applications.
|
||||
- Fix to errorIndex generation at CommandResponder application, the value
|
||||
should be a one-based.
|
||||
|
||||
-------------------------------------------------------------------
|
||||
Mon Apr 23 2007 James Oakley <jfunk@funktronics.ca> - 4.1.7-1
|
||||
|
||||
- Update to 4.1.7
|
||||
- Low-level debugging facility implemented
|
||||
- Support UdpTransportTarget timeout and retries parameters in oneliner API
|
||||
- Fix to snmpTrapOID construction at ...proxy.rfc2576.v1ToV2()
|
||||
function
|
||||
- Fix to MibViewController.getNodeName() to take MIB module name
|
||||
into account (SF bug #1505847)
|
||||
- Do explicit check for Counter32,Unsigned32,TimeTicks,Counter64 value types
|
||||
in MibTableRow index convertion and in TextualConvention.prettyPrint()
|
||||
methods (SF bug #1506341). Handle Bits in indices as RFC2578 suggests
|
||||
- Apply read-create column status to libsmi2pysnmp-generated code
|
||||
whenever MIB text specifies that (SF bug #1508955)
|
||||
- Honor and apply DISPLAY-HINT specification when building TextualConvention
|
||||
class at libsmi2pysnmp
|
||||
- Managed Objects Instances files (smi/mibs/instances/) are now
|
||||
double-underscore prefixed to make them imported explicitly by these
|
||||
prefixed names. They used to be imported as a side-effect of
|
||||
Managed Objects files import what is way too hackerish
|
||||
- The libsmi2pysnmp now supports future libsmi bugfix that would generate
|
||||
"ranges" subtree along with the legacy and ambiguous "range" one
|
||||
- SMI support for fixed-length string indices implemented (SF bug #1584799,
|
||||
- Managed Object Instances may now have smiRead, smiWrite, smiCreate methods
|
||||
to support specific value mangling. These methods, if present, would be
|
||||
invoked from SNMP [Agent] core instead of conventional clone() method.
|
||||
The reason is to separate trivial value duplication from specific
|
||||
Instance value mangling that may have Agent-specific side effects
|
||||
(such as RowStatus)
|
||||
- MIB table row destruction now works (SF bug #1555010)
|
||||
- LCD unconfiguration functions for oneliners implemented (SF bug #1635270)
|
||||
- unloadModules() and unexportSymbols() implemented at MibBuilder
|
||||
- Notification type PDU proxy code fixed to produce symmetrical
|
||||
convertion
|
||||
- Various SNMP engine-internal caches expiration implemented
|
||||
- SMI-level access control now takes effect only if AC object is
|
||||
passed to MIB instrumentation API
|
||||
- LCD management code now uses generic MIB instrumentation features
|
||||
- Fix to oneliner manager code to have individual UdpSocketTransport
|
||||
instance per each SnmpEngine. Multithreaded apps might fail otherwise.
|
||||
(SF bug ID #1586420)
|
||||
- Exclude the PYSNMP_MODULE_ID symbol from MIB view index, as it may get
|
||||
resolved into OID label instead of actual MIB object name
|
||||
- Memory leak fixed in indices.OidOrderedDict implementation
|
||||
- Fix to VACM shortcuts to let notifications and requests to co-exist
|
||||
for the same user otherwise
|
||||
- Fix to ...oneliner.cmdgen.UsmUserData to support non-default cyphers
|
||||
- USM now uses local notion of snmpEngineBoots/Time when authoritative
|
||||
and cached estimate otherwise. Also, a security fix applied to to USM
|
||||
time-window verification (SF bug #1649032)
|
||||
- Fix to notification objects resolution code at
|
||||
NotificationOriginator.sendNotification()
|
||||
- Do not raise securityLevel for USM error reports that lacks user
|
||||
information, as these reports could never be cyphered (SF bug #1624720)
|
||||
- Non-default BULK PDU params now actually applied
|
||||
- SnmpEngineID default value generation algorithmic function changed
|
||||
to allow multiple SNMP engines running on the same host
|
||||
- Documentation updated
|
||||
- A handful of minor fixes applied (SourceForge tracker IDs #1537592,
|
||||
|
||||
-------------------------------------------------------------------
|
||||
Thu Jun 22 2006 James Oakley <jfunk@funktronics.ca> - 4.1.6-1
|
||||
|
||||
- Update
|
||||
|
||||
-------------------------------------------------------------------
|
||||
Wed Oct 19 2005 James Oakley <jfunk@funktronics.ca> - 4.1.4-ft.4
|
||||
|
||||
- Fix SNMPv1 requestid problem
|
||||
|
||||
-------------------------------------------------------------------
|
||||
Wed Oct 12 2005 James Oakley <jfunk@funktronics.ca> - 4.1.4-ft.3
|
||||
|
||||
- Build for SL10
|
||||
|
||||
-------------------------------------------------------------------
|
||||
Wed Sep 28 2005 James Oakley <jfunk@funktronics.ca> - 4.1.4-ft.2
|
||||
|
||||
- Add requirements
|
||||
|
||||
|
Reference in New Issue
Block a user