Difference between revisions of "ChucK/Dev/Shell"

From CSWiki
Jump to: navigation, search
(documentation outline)
(documentation outline)
Line 28: Line 28:
***(mention --loop behavior)
***(mention --loop behavior)
* [[ChucK/Dev/Shell/doc#Command_Summary|shell commands ]] (see below)
* [[ChucK/Dev/Shell/doc#Command_Summary|shell commands ]] (see below)
* in-shell coding
* [[ChucK/Dev/Shell/doc#Inline_Coding|in-shell coding]]
* option tutorial 2
* option tutorial 2
[[ChucK/Dev/Shell/cmd|shell commands]]

Revision as of 00:57, 1 March 2006


  • shell
    • support intended functionality
      • support on-the-fly programming
      • execute code inline
      • run-time code modification
    • framework component re-use


  • have a shell toggled by flag (--shell)
  • interfaces classes with document
  • on-the-fly programming
  • checkpoint (11/14)
  • shell documentation
  • miniAudicle documentation

documentation outline



  • --shell automatically starts a VM
  • shell exits when VM is finished running currently, but keeping the shell alive after the VM ends is an option to consider
  • --empty restricts VM
  • chuck @ hostname -pPORT --shell causes ChucK shell to automatically attach/add the VM at hostname:port. --empty is implicit in this option
  • shell commands
    • vm attach [hostname[:port]] or vm @ [hostname[:port]] attempts to attach to a ChucK VM running on hostname:port. The default port is 8888. If no hostname or port is specified, the shell attempts to attach to the in-process VM
      • prints a success message and the VM reference number for each successful attachment; prints a failure message for each failed attachment
    • vm + or vm add
      • saves the current VM for later use
    • vm - [vm_reference] or vm remove [vm_reference]
      • removes specified VM--it is no longer on the VM list
    • vm = [vm_reference] or vm swap [vm_reference]
    • vms or vm list
      • lists all VMs plus the current VM
    • vm @+ hostname:port adds and attaches the specified VM
    • add [files] or + [files]
      • adds [files] to current VM and prints out a shred reference number for each file
    • remove [ids] or - [ids]
    • removeall
    • replace id filename [id filename]...
    • -- (remove shred with highest id)
    • status or ^
    • kill kills the current vm--shuts down the shell too if current_vm is the process vm
    • ls (like unix ls)
    • alias and unalias (simple substitution-based, bash-like command aliases)
    • shreds list all shreds on the current VM
    • pwd
    • cd (change directory, also change current dir for chuck so paths are consistent)
    • exit exit everything
    • close close down shell, keep process VM running
  • Ge: thoughts on these commands? ^
    • they look great! few questions/thoughts:
      • depending on the command line flag issued, the shell could automatically attach the most appropriate thing
      • (maybe by default, the shell is attached to 'process'
      • --local and --process probably don't need to be be flags since they are targets of the attach shell command (local or process or self may suffice?)
      • the multiple attached target vms is totally cool!

executing chuck code dynamically

  • { [code] } adds code to the current VM
    • A { } block can be stretched out over a number of lines
    • if so, the shell prompt will change to code and will also print the current scope level of the code entry. Also, the code entry will be considered complete when the shell finds a matching brace for the first opening brace
    • If an alphannumeric text string is supplied after the final brace, the code will be saved within the shell and will be associated with that particular name. Non-CR/LF whitespace may separate the brace and the code name.
      • example: { 0 => int i; } foo will be saved as 'foo'
    • (potential advanced feature) specifying an arbitrary alphanumeric context name after the keyword context will apply the state context of that name to any code segment which opens on that line.
      • example: context bar { 0 => int i; } will apply context 'bar' to the following code
      • if a context doesn't exist, it will be created using the state of the subsequent code entry
      • a context is essentially a scope. by executing two segments of code in a single context, they will have access to the same top-scope variables
      • if no context is specified, then no new context is created and no special context is used
  • code list [code_name]
  • code print [code_name]
  • code save code_name saves last code entry as code_name
  • code delete code_name
  • code add code_name adds code_name to current VM
  • code write code_name file writes code_name to file
  • code read code_name file reads code in file to code_name

things that would be nice to have from existing chuck code

  • more feedback from otf_ functions
    • status sends status report back to calling process
    • add sends back the shred number of the added process
  • single exit path for Chuck_VM -- instead of calling exit(), it would be helpful if the VM simply stopped running when it recieves a kill command