Ncat-Lua/Meeting agenda

These are my personal notes for my upcoming meeting with David.

27th August 2013

• --lua-exec documentation
  • remember to mention infinite loops leading to leaving alive child processes (did I fix that already?)
• connect mode and blocking reads
• close not disconnecting cleanly ("broken pipes") - BTW, mention the vanilla ncat -l ^C not disconnecting cleanly either
• close in responder.lua, connect mode not saying bye before closing.
• double close leading to segfault - a workaround was to set userdata to NULL to make assertions fail.
• error behavior

3rd September 2013

• raw sockets?
• HACKING file for Ncat?
• [forgot to talk through the main thread last time] close not disconnecting cleanly ("broken pipes") - also, I thought close() right after accept() should refuse the connection, why doesn't it?
• echo while true do end > crash.lua ; ncat --lua-exec crash.lua -l & sleep 1 ; ncat localhost - should it die on disconnection? (POSIX only)
  • not really Lua-specific - could reproduce it under Python too. Still - it's inconsistent, why does it work this way?
• I made a bit a mess of during experimenting with Secwiki - ask David to remove the two pages from Category:Pages for deletion
• BUG: compare ncat -l > result.txt on Windows on Unix, --lua-exec'ing the following script on the client side:

io.stdout:write "A"
io.stdout:write "\r"
io.stdout:write "B"
io.stdout:write "\r\n"
io.stdout:write "C"
io.stdout:write "\n\r"
io.stdout:write "D"
io.stdout:write "\n"
io.stdout:write "E"

10th September 2013

• which branches am I safe to delete?
• ncat -l --sh-exec "tee /dev/stdin" & ncat localhost, then type in something - why do I get spammed?
• luaL_ref vs connections[fd] (got no e-mail reply)
• problems with stopping recv/send if Lua returned nil