Import release 0.1.12

[secnet.git] / TODO

diff --git a/TODO b/TODO
index af3a4686efd4ad5be8ee2a6f870b4f78aaa48323..c7d82489dffabbc343dfc13b854ddd223a43caa2 100644 (file)
--- a/TODO
+++ b/TODO
@@ -1,20 +1,14 @@
-Makefile.in: autodep stuff
-
 dh.c: change format to binary from decimal string (without introducing
 endianness problems)
 
-netlink.c: investigate why 'default' routes don't appear to work
-(reported by JDA).
+netlink.c: test the 'allow_route' option properly.
 
-slip.c: detect failure of userv-ipif to start. Restart userv-ipif to
-cope with soft routes? Restart it if it fails in use?
+process.c: capture output from children in sys_cmd() and log it
 
-tun.c: jdamery reports tun-old code works on Linux-2.2.
-Unresolved problem with ioctl(TUNSETIFF) sometimes returning EINVAL, seems
-to be related to early 2.4.x (x<=5) series kernels. 2.4.9 and above seem ok;
-2.4.[678] untested.
+random.c: test properly
 
-random.c: test
+resolver.c: ought to return a list of addresses for each address; the
+site code ought to remember them and try contacting them in turn.
 
 rsa.c: check padding type, change format to binary from decimal string
 (without introducing endianness problems)
@@ -22,16 +16,15 @@ rsa.c: check padding type, change format to binary from decimal string
 site.c: the site_incoming() routing could be implemented much more
 cleanly using a table. There's still quite a lot of redundancy in this
 file. Abandon key exchanges when a bad packet is received. Modify
-protocol to include version fields, as described in the NOTES file.
-
-transform.c: see below
-
-sha1.c: test
-
-General: separate the transforms in transform.c into multiple parts,
-which can then be combined in the configuration file.  Will allow the
-user to plug in different block ciphers, invent an authenticity-only
-mode, etc.
-
-Signal handling! Really just cope with SIGCHLD and SIGTERM. Possibly
-use SIGUSR1/2 for prodding things. Manage child processes properly.
+protocol to include version fields, as described in the NOTES
+file. Implement keepalive mode. Make policy about when to initiate key
+exchanges more configurable (how many NAKs / bad reverse-transforms
+does it take to prompt a key exchange?)
+
+slip.c: restart userv-ipif to cope with soft routes? Restart it if it
+fails in use?
+
+transform.c: separate the transforms into multiple parts, which can
+then be combined in the configuration file.  Will allow the user to
+plug in different block ciphers, invent an authenticity-only mode,
+etc.