4 This is more or less complete list of tasks that has to be done before
5 SILC 1.0 could ever be released. It is clear that the list does not
6 include all the bugs that exists. At the end of list are tasks that
7 needs to be done but are probably post 1.0.
9 Feel free to contribute if you have the ability and free time - all the
10 help is really appreciated - and needed.
18 o Optimization in general. I have not focused to program optimized code
19 in many circumstances. It has been more important to get this up and
20 working. A lot must be optimized especially the ID cache. Instead of
21 using lists some hash tables should be used. The searching by ID and
22 such must be made a lot faster. If someone would like dedicate their
23 efforts purely to generate optimized code I'd be appreaciated.
25 o We should replace all short, int, long, unsigned short, unsigned int,
26 unsigned long with some pre-defined datatypes that really are what
27 we want on all platforms. int16, uint16, int32, uint32 etc. are
28 what we could use or maybe SilcInt16, SilcUInt16 etc. Also, boolean
29 datatype should be defined.
31 o Add boolean (or bool), typedef of unsigned char.
33 o More platform supports should be added. The code is pretty much
34 generic but there are some parts that require porting (SIM). Also,
35 some support for different platforms is needed into configure.in.
37 o SILC requires currently GCC to work because we use GCC specific
38 compilation options. Generally any compiler that supports inline
39 functions and can build shared libraries (for SIMs) should work.
40 These cases should be included into configure.in.
42 o Extended SIM (SILC Module) support. Currently only SILC Cipher API
43 and SILC Hash API may be used as SIM's. What I have in mind is to
44 have extended support for SIM's so that basically any SILC API could
45 be used as SIM's. This would open tremendous possiblities but
46 opens also issues on security that needs to be dealt with.
48 Some sort of SIM compilation environment should be defined so that
49 the SIM's could use SILC specific symbols from the modules (which they
50 cannot do currently). In the future modules could add new features
51 to SILC easily with this support. I'm more thinking this from client's
52 perspective to add new features to client (such as IRC support as SIM)
53 but server should have the support as well. Anyhow, this is an
54 interesting feature...
56 This maybe post 1.0 task - dunno.
58 o SIM support for other platforms than just for Linux. Apache has
59 example code (code that we could use directly pretty easily) for
62 TODO In SILC Client Library
63 ===========================
65 o Non-blocking connection on the background must be stopped if some
66 other connection on same window has established. Now it is possible
67 that some non-blocking connection timeouts on the background when
68 we already have a working connection to some other place; things
71 o Input line on UI is buggy. Cursor movement etc bugs. Too lazy to
74 o Logic for handling multiple same nicknames for example in private
75 message sending. I guess the logic is done in server side but is
78 o Private message key setting is missing and must be implemented.
79 Currently private messages are encrypted with session keys. This
80 is required by the protocol.
82 o Channel private key setting is missing and must be implemented.
83 Currently there cannot be private keys for channels. Normal channel
84 keys (generated by server) are used. This is required by the protocol.
86 o I guess, public key authentication (when connecting to a server)
87 is not working currently. It is just matter of loading the keys
88 from file and using them (see corresponding code in server, it should
89 support public key authentication already).
91 o Connection Authentication request resolving is missing and must be
92 done. This is required by the protocol.
94 o Move ssh_client_notify_by_server to its own file (like notify.[ch]).
96 o Key Exchange protocol's responder side is missing from client.
97 Generally it is possible for the client to be responder so it should
98 be implemented (See corresponding code from server). Error handling
99 in the KE protocol is also in pretty bad shape in client.
105 o DNS/IP lookup blocks the server. This must be fixed. Check the
106 resolver stuff (resolver(3), resolver(5)). Either we have to do the
107 own resolver stuff (through scheduler, if possible without writing
108 too much own stuff) or use threads.
110 o Length of the packet processing timeouts needs to be checked whether
111 they are too short or too long. I haven't really tested whether they
112 are suitable. They should be tested on high load which I haven't done
115 o INVITE command must set the channel's invite list if channel is
118 o Server says that it is able to listen on multiple ports but currently
119 that is bogus. It can, but internals are for single server.
121 o Command flag usage in general is not implemented yet.
123 o Client history must be implemented. Protocol says that server must
124 keep history information about clients for some period of time.
126 o Protocol execution timeouts are hard coded, should be configurable.
128 o serverconfig.c and the function naming in it is inconsistent. It is
129 not silc_config_server* it should be silc_server_config*. As should
130 all the SilcConfigServer* types be SilcServerConfig*.
132 o Implement DENY_CONNECTION section in serverconfig.c and in server.
134 o Implement REDIRECT_CLIENT section in serverconfig.c and in server.
136 o Configuration file format - could be better.
138 o IP address fields in configuration file should accept mask format
139 as well, IP/MASK, and not just plain IP.
141 o Connection classes should be actually implemented in serverconfig.c.
142 They can be defined but they are totally ignored currently.
144 o Acceptance of incoming connections (client and server connections)
145 should be checked before key exchange protocol. Currently it is
146 checked at the authentication phase after KE, that is ok, but it should
147 be checked before starting KE, as well.
149 o Statistics are totally missing from the server. It would be nice
150 to gather some statistics.
153 TODO In SILC Libraries
154 ======================
156 o Rewrite the task system. I made it too complex and too "neat" and
157 it really should be rewritten. We don't need priorities really, one
158 priority is enough. This will simplify a lot the task system.
160 o Implement PFS (Perfect Forward Secrecy) flag in SKE (and in client and
161 server, actually). If PFS is set, re-key must cause new key exchange.
162 This is required by the SILC protocol.
164 o silc_id_str2id must also take ID length as argument. Otherwise, variable
165 length ID's (after we add IPv6) will not work.
167 o Re-key in general is actually missing (from everywhere) and must be done.
169 o ID Cache expiry does not work.
171 o PKCS#1 style RSA public key encryption/decryption/sign/verify is
172 missing, and should be added for interoperability reasons. The thing
173 I've done now is bad and should be removed as soon as possible (or
174 the protocol should then state the method of how they should be done).
176 o Scheduler needs to be analyzed on high load as it might be unfair
177 towards select() because it may run timeout tasks before select() and
178 after select(). If it is found to be unfair the timeout task running
179 before select() should probably be removed.
181 o On select() issue; maybe we should use poll() instead if it is
182 available? poll() doesn't have max fd limit...
184 o SIM support for SILC PKCS API needs to made so that they could be
185 used as SIM's. At the same time some work is required on prime
186 generation as the way it is done now sucks. Read from code for
189 o Compression routines are missing. The protocol supports packet
190 compression thus it must be implemented. SILC Comp API must be
191 defined. zlib package is already included into the lib dir (in CVS,
192 not in distribution), but it is not used yet, and it requires some
193 tweaking on the Makefiles (we want static lib not shared).
195 o Cipher API needs to be made more consistent. Some parts of the
196 code generated with current Cipher API looks really bad. Same
197 is with PKCS API, even worse actually. They need to be made
198 cleaner. Introducing silc_cipher_encrypt/decrypt/set_key etc.
199 functions (I actually don't understand why have I left these un-done).
201 o Random Number Generator needs some tweaking. Reading /dev/random may
202 block resulting slow initialization of RNG. Some other things in the
203 RNG may block as well. Also, I have some pending changes to the RNG
204 that needs to be commited (from Schneier's Yarrow-160 paper). They
205 should make the RNG even better.
207 o silc_buffer_[un]format() needs to be made more stable as it may
208 crash the SILC if malformed data is sent as argument. There are a
209 lot of places in client and server where we trust directly data coming
210 from network and try to unformat it. The unformatting routine needs
211 to be able handle situations where data sent is malformed, by mistake
212 or intentionally. This is important as it is easy to crash the SILC
213 now by just sending malformed data. Also, in client and server we
214 must start checking the return value from silc_buffer_[un]format.
220 o Write manuals for server.
222 o Write manuals for client.
224 o Write SILC Library Reference manual. This would include all the SILC
225 API's with simple examples how the functions are to be used. This is
226 pretty easy to create by taking all the functions plus their comments
227 from source/header files. However, same sort of reference manual
228 should be written for client and server as well.
234 o Pthreads support. A lot of problems are solved with server (and with
235 client as well) if we add pthread support. We can forget things such
236 as non-blocking connecting etc, and we can do things such as DNS/IP
237 lookups async. The server itself also benefits great deal from
238 threads, especially from performance point of view.
240 But, this is not a small task and almost entire SILC Library has to
241 be made re-entrant. Own API is probably added for the threads support
242 to make changes in the future as painless as possible. So the API
243 would have things like silc_mutex_lock, silc_mutex_unlock and
246 o X.509 certificate support. SILC protocol supports certificates and
247 it would be great to have support for them. This is a big task as
248 support has to be made for ASN.1 as well. I've looked into OpenSSL
249 package as it has X.509 certificate support (and ASN.1 as well).
250 The code does not look very good to my eye but it has some potentials.
251 This should be looked at more closely.
253 Naturally own SILC Certificate API has to be defined regardles what
254 the actual X.509 library is (OpenSSL X.509 or something else). Other
255 choice is to write own X.509 library but I'm not going to do it -
256 I can help to migrate the OpenSSL X.509 into SILC and I can help if
257 someone would like to write the X.509 library - but I'm not going
258 to start writing one myself. Anyhow, the OpenSSL X.509 lib should
261 o SSH2 public keys support. Maybe - not really needed but could be
262 nice as SSH is widely used all over the place. SILC Protocol
263 supports SSH2 public keys.
265 o Cipher optimizations (asm, that this) at least for i386 would be nice.