Clojurians Log v2

Calva’s nREPL log has this when it happens:

-> sent
{
  op: 'complete',
  ns: 'user',
  symbol: 'igno',
  id: '95',
  session: '9b241d55-c918-43fa-b776-19b41a80bc43',
  context: '{:ignr __prefix__\n' +
    ';; On ^that^ line:\n' +
    ';; 1. continue typing one letter after the word starting with `ign`\n' +
    ';; 2. hit backspace (assuming you have strict structural delete enabled)\n' +
    ';; 3. the letter is deleted.\n' +
    ';;    Try typing some more and backspace over it.\n' +
    ';;    It works.\n' +
    ';; 4. Start and connect the REPL\n' +
    ';; 5. Goto 1.\n' +
    ';; 6. Instead of 3. experience lockup,\n' +
    ';;    The letter will eventually be deleted\n' +
    '\n' +
    '          (?:                      # Ignorables\n' +
    '            \\#\\!.*\\n? |              # hashbang line\n' +
    '            [\\s,]+    |              # whitespace, commas,\n' +
    '            ;.*\\n?                   # comments\n' +
    '          )"\n' +
    '    :char #"\\\\."                   ; Character token\n' +
    '    :keyw #"(?:\\:\\w+(?:-\\w+)*)"    ; Keyword token\n' +
    '    :oper #"[-+*/<=>|&]{1,3}"      ; Operator token\n' +
    '    :str  #"(?x)\n' +
    '          \\#?                      # Possibly a regex\n' +
    '          \\"(?:                    # Quoted string\n' +
    '            \\\\. |                    # Escaped char\n' +
    '            [^\\\\\\"]                  # Any other char\n' +
    '          )*\\"?                      # Ending quote\n' +
    '          "\n' +
    '    :sym  #"\\w+(?:-\\w+)*[?!]?"     ; Symbol token\n' +
    '    :symp #"\\w+(?:-\\w+)*[?!]?\\("   ; Symbol followed by paren\n' +
    '    })\n' +
    '\n' +
    ' (defn re [s]\n' +
    '   (re-pattern\n' +
    '     (reduce\n' +
    '       (fn [re [k v]]\n' +
    '         (let [pat (re-pattern (str #"\\$" (subs (str k) 1) #"(?!\\w)"))]\n' +
    '           (str/replace re pat (str/re-quote-replacement v))))\n' +
    '       s dict)))',
  'enhanced-cljs-completion?': 't'
}

<- received
{
  id: '94',
  list: [],
  session: '9b241d55-c918-43fa-b776-19b41a80bc43',
  status: [ 'done' ]
}

<- received
{
  completions: [],
  id: '95',
  session: '9b241d55-c918-43fa-b776-19b41a80bc43',
  status: [ 'done' ]
}

I haven’t tried it with CIDER, yet, but I’m guessing it might not send that broken code as context. I will see when I try it. Or someone else will see, beating me to it, because I won’t be doing it tonight.

OK. After having given this some more thought, I don’t think I have enough data to conclude this happens in nREPL. It could be that it is Calva that struggles with collecting the context. I think I should start with adding time stamps to those nREPL message log entries… Anyway, I’ll be looking at it from the Calva end first, and will update about wether I have reason to think something needs to be looked at in the nREPL server.

Indeed. With timestamps on the messages I can clearly see that there is nothing getting stuck in nREPL. This is all going wrong in Calva. Sorry about the noise!