Arc Forumnew | comments | leaders | submitlogin
Reloading .arc files inside the web server
3 points by hasenj 4878 days ago | 6 comments
When running a web server with (asv), is there a way to reload the server without terminating the arc process?

Can I make it reload .arc files automatically when they're changed?



4 points by evanrmurphy 4878 days ago | link

If you run the arc server as a thread, you can load in files while it's still running or restart the server without killing the arc process:

  ; Start the arc server in a thread called t*

  arc> (= t* (thread (asv)))
  #<thread: t*>
  arc> ready to serve port 8080

  ; Load a file while the server is still running  
  ; Every function and macro in arc.arc is redefined (ellipsized here for brevity)

  arc> (load "arc.arc")
  *** redefining caar
  *** redefining cadr
  *** redefining cddr
  ...

  ; Stop the server without terminating the arc process by
  ; breaking the thread

  arc> (break-thread t*)
  #<void>
  arc> user break

   === context ===
  /home/evanrmurphy/programas/arc/arc3.1/ac.scm:1040:20: socket-accept
   handle-request-1
   gs1046
  zz

  ; Restart the server by creating the thread anew

  arc> (= t* (thread (asv)))
  #<thread: t*>
  arc> ready to serve port 8080

-----

2 points by hasenj 4877 days ago | link

Thanks Evan, nice trick!

Now, can we watch filesystem for changes?

I suppose we can sleep and poll the timestamp, but I'm running into a problem, mtime is defined in lib/files.arc,

The way I'm settings things up, my experimental "app" is in a completely isolated folder, and once I'm in there, "lib/files.arc" is no longer in the current directory, how should I go about loading stuff from "lib/"?

For now I only need the 'mtime' function so I just snatched it, but I hope there's a better way of importing standard libs.

Otherwise, here's a little file watcher I just wrote, seems to work with my tiny test case:

    (def mtime (path)
      " Returns the modification time of the file or directory `path' in
        seconds since the epoch. "
      ($.file-or-directory-modify-seconds path))

    (= file-ts* (obj))
    (def file-ts-changed (file-name)
        (no:is (file-ts* file-name) (mtime file-name)))
    (def update-file-ts (file-name)
         (= (file-ts* file-name) (mtime file-name)))

    ; watch file system for updates to file-name
    ; and run 'delegate' when file is updated
    (def watch (file-name delegate)
         (update-file-ts file-name)
         (while (no (file-ts-changed file-name))
                (sleep 2))
         (delegate)
         (update-file-ts file-name))

    (def watch-load (file-name)
         (repeat 2 (prn))
         (pr "watching " file-name " for changes ..")
         (watch file-name 
                (fn() (load file-name) (pr "reloaded " file-name))))

    (watch-load "web.arc")

-----

1 point by bogomipz 4877 days ago | link

Just a couple of comments on the watch function:

1. Shouldn't the last line be (watch file-name delegate) so that it continues to watch after the first update? Or should that be optional?

2. It might be a good idea to pass file-name to the delegate. Not strictly necessary, and doesn't give any benefit in this example, but it could mean that you sometimes don't need to use a fn as the delegate. If you didn't want to pr "reloaded" for instance, you could just do (watch file-name load).

3. Is the hash table really necessary? Here is a version without it:

  (def watch (file-name delegate)
       (let ts mtime.file-name
         (while (is ts mtime.file-name)
                (sleep 2))
         (delegate)))

-----

1 point by hasenj 4877 days ago | link

I put the call to watch as the last line inside the file itself.

And the hash table is there because I initially wanted to watch multiple files at the same time. But I suppose if the main file in the web app loads the other files, then it's not needed at all.

The other thing I did is running the (asv) thread only once by checking for an unbound symbol, so that when the file is reloaded it doesn't run the server again.

  (def srvr ()
    (if (no (bound 't*))
      (= t* (thread (asv)))))

  (srvr)

-----

1 point by bogomipz 4875 days ago | link

Ah, so loading the file makes it watch itself. You need to do the load in a thread then?

I still fail to see how the hash table helps since my version above should do exactly the same as yours. I would see the point if you made a thread check all files in the table, and made the first call to watch launch the thread, while subsequent calls just add entries to the table.

-----

3 points by hasenj 4875 days ago | link

Well actually, I'm putting the (asv) call in a thread and making it so that reloading the file doesn't restart the server:

  (mac set-once (s v)
       `(if (no:bound ',s)
          (= ,s ,v)))

  (set-once t* (thread:asv))
Watching for changes is not running in a thread, it's just the last line in my "web.arc"

And you're right, the hash-table is not needed at all. That was a premature design.

I actually re-wrote it a bit:

  ; watch the result of (func) for change
  ; and run 'on-change' when value changes
  (def watch-fn (func on-change)
       (let init (func)
           (while (is init (func))
              (sleep 2))
       (on-change)))
  
  ; watch the value-exp expression for changes
  ; and run on-change-exp when it changes
  (mac watch (value-exp on-change-exp)
       `(watch-fn (fn() ,value-exp) (fn() ,on-change-exp)))
  
  (def watch-file (file-name deleg)
       (watch (mtime file-name) (deleg file-name)))
  
  (def auto-reload (file-name)
       (prn "w/auto-reloading " file-name)
       (watch-file file-name load))
  
  (auto-reload "web.arc")

-----