Google Git
Sign in
gfiber / vendor / opensource / netsnmp / 8934b68cb96fa4807430aeb93803e2361a5707ed / . / TODO
blob: 8f0ecd381e9ae398ea0ea6b6654c5531b3118718 [file] [log] [blame]
Wes Hardaker56882591997-02-04 08:35:37 +0000[diff] [blame]1TODO list for the ucd-snmp-coders:
2
Wes Hardaker387ec511998-05-29 22:20:53 +0000[diff] [blame]3 possibly add WinSNMP wrappers around the ucd-snmp functions
Wes Hardaker873ae8b1997-02-05 15:33:32 +0000[diff] [blame]4 Continue to add and improve manual pages/documentation.
5 More snmpset'able variables.
6 Should be able to set the fix script line in the .conf files per type.
Wes Hardaker873ae8b1997-02-05 15:33:32 +0000[diff] [blame]7 Implement IP routing table manipulation using sysctl() on systems that
Wes Hardaker35bf1961997-05-09 20:46:46 +0000[diff] [blame]8 support the route socket (BSD 4.4 and IRIX >= 6.2)
Wes Hardaker873ae8b1997-02-05 15:33:32 +0000[diff] [blame]9 Implement new IP forwarding table MIB (RFC 2096), at least on systems
10 that have classless routing tables internally
11 Implement IP/UDP/TCP statistics on IRIX 6.2 and later using
12 sysmp(MP_SAGET,MPSA_TCPIPSTAT,...) and <sys/tcpipstats.h>
13 Follow the evolution of the Interfaces Group MIB
14 (draft-ietf-ifmib-mib-05.txt) and implement what can be
15 implemented.
16 Make read_objid and sprint_objid agree on which shorthands they accept.
Wes Hardaker4cd67fd1997-05-20 14:50:18 +0000[diff] [blame]17 snmpset should check against MIB size if possible?
Wes Hardaker9f9266a1999-02-03 06:01:01 +0000[diff] [blame]18 Remove unsuppored mib nodes from output instead of returning '0's.
Niels Baggesenac47ddc1999-06-22 12:25:13 +0000[diff] [blame]19 Support for Counter64 in snmpdelta
Wes Hardaker7f91fdf2000-02-05 01:49:46 +0000[diff] [blame]20 API for traps that pass scripts can use.
Wes Hardaker1b1708e1999-11-15 23:19:47 +0000[diff] [blame]21 Make memory/swap variables UInts instead of Ints.
Wes Hardaker7da65d91995-10-03 16:24:39 +0000[diff] [blame]22
Wes Hardakera934a2e1998-06-10 22:30:59 +0000[diff] [blame]23===================
24
25Issues:
26
27 CODE-RELATED ISSUES
28 -------------------
29
30Architecture-specific implementations:
31 How to handle differences between architectures
32 - overall common template
33 - "standard" implementation (with minor differences)
34 - architecture unique implementations
35 pre-processor #ifdefs or separate files
36
37Function argument handling:
38 Function declaration prototypes
39 Function definition prototypes
40 How to handle both ANSI and K&R compilers
41
42Threading
43 Single thread / multi-thread / event driven / asynchronous
44
45Lookup design
46 Kernel digging vs "clean" interfaces (sysctl/ioctl)
47
48Error Handling
49 Ensure all return values are checked for validity
50
51Documented APIs
52
53WinSNMP interface
54
55Package name
56 UCD-snmp vs Net-snmp
57
58
59
60 FUNCTIONALITY-RELATED ISSUES
61 ----------------------------
62
63Per-interface information
64 (speed/type/counters/non-ethernet address)
65
66Multi-CPU support
67
68Real Time O/S support
69
7064-bit counters
71
72Non-ethernet interface support
73 Address handling code tends to assume 6 octet
74 physical addresses and 4 octet transport addresses
75
76Non-fully-qualified name output
77 (Probably covered now?)
78
79Formatted dump
80 unpick the ASN encoding, rather than
81 simply dumping the bare data
82
83Module interdependence
84
85
86
87 EXTENSIBILITY-RELATED ISSUES
88 ----------------------------
89
90AgentX
91
92SMUX
93
94Dynamic loading of modules
95
96"Intelligent" Agent
97 agent-initiated traps
98 threshold monitoring
99
100PHP scripting
101
102Extensible trapd
103
104
105
106 PERFORMANCE-RELATED ISSUES
107 --------------------------
108
109Memory Leaks
110
111Information caching
112 (e.g. Host Resources information)
113
114Intelligent table handling
115 (i.e. not resetting internal state for each request)
116
117Pre-compiled MIBs
118 to speed up MIB parsing
119
120Duplicate requests
121 Currently these are processed afresh each time.
122 They should probably be either dropped or use cached replies
123
Wes Hardakera6948be1998-09-03 01:43:30 +0000[diff] [blame]124Security
125 Go through the library and apps and check all arrays and
126 buffers for overruns. There are a lot of unchecked ones.
Wes Hardakera934a2e1998-06-10 22:30:59 +0000[diff] [blame]127
128===================
129
Wes Hardaker56882591997-02-04 08:35:37 +0000[diff] [blame]130Other Ideas:
Wes Hardaker04ca0331996-08-26 08:37:05 +0000[diff] [blame]131 Make the mib parser figure out which of the above files it needs to
132 load rather than loading them all -- Tom Georges & Joe Marzot
Wes Hardaker04ca0331996-08-26 08:37:05 +0000[diff] [blame]133
Wes Hardaker3a461bc1995-10-16 22:55:37 +0000[diff] [blame]134===================
135
136Want something done? Have ideas? Have code to submit? Need
Wes Hardakera3008922000-12-02 00:34:23 +0000[diff] [blame]137something earlier than it is listed above? Please!
138 Write us!
139 net-snmp-coders@lists.sourceforge.net
Wes Hardakera934a2e1998-06-10 22:30:59 +0000[diff] [blame]140
141(even if just to announce that you are going to be working on it so we
142can avoid multiple people working on the same thing)