1 May 1998: The _COMMAND add-on for SRE-http


The _COMMAND add-on for SRE-http is used to remotely issue (through
a web browser) OS/2 commands to your server. These can be ANY OS/2 command,
as long as it does not require user input. Alternatively, you can "submit"
REXX programs to your server.

   You can think _COMMAND as giving SRE-http a "batch Telnet" capability.

Setting up and using  _COMMAND is simple:

  1)Unzip _COMMAND.ZIP to a temporary directory
  2)Copy _COMMAND.CMD to your SRE-http "addon" directory (where you've
    installed SRE-http's configuration and add-on files),
    i.e.; D:\GOSERVE\ADDON)
  3)Copy _COMMAND.HTM to your GoServe data directory.


Running _COMMAND:

  *   The simplest way of using _COMMAND is to point your browser at
      _COMMAND.HTM.  You will be given two choices: entering an OS/2 command,
      or submitting a REXX program (which you write on the spot).  In both
      cases, the output of the command (or program) will be returned to your
      browser.  This includes most (but not all) error messages.

  *   The "submit program option" is meant for small REXX programs that you
      write in a TEXTAREA on your browser. It should be a  syntatically
      correct REXX program (errors will be trapped, with a simple error
      message returned).
      There are a few  important points to consider when "submitting
      REXX programs":
          1) Do NOT USE GoServe commands (such as EXTRACT) -- you WILL
             CRASH GoServe if you do!
          2) To output results to the client (yourself), use the QUEUE
             command. For example:
                  QUEUE  'The time is now ' time('n')
 
  *   If your command (or REXX program) requires a long time to complete,
      GoServe might time out. If this happens, SRE-http will attempt to
      clean up the various pieces-- but there is a slight chance that the
      server will be slowed down (until the process you invoked is finished).
 

  *   Feel free to modify _COMMAND.HTM -- the syntax is simple.
      For example, you can have several different commands submitted
      per request.

  *   Note that the "default" directory for _COMMAND is the
      GoServe working directory (where you installed SRE-http)

  *   If you like entering URL's by hand, you can make the following types
      of requests:
           http://your.server/_command?type+foo.bar
      where "type" can be any OS/2 command. Note the use  of + instead of
      space characters (HTTP does not tolerate spaces in the request string).
      One nice trick is to issue two commands, by using the & character.
      For example:
           http://your.server/_command?copy+\www\index.htm+index.v1&dir+index.*

   *  As with the DETACH command, you should only use this with commands
      that do not require further user input (i.e.; del d:\foo\bar will
      ask for a Yes/No response -- this response which will be echoed back
      to the client)



 
A Note on Security:

  Obviously, the _COMMAND utility can be disasterously abused by nefarious
  individuals.  By default, only SUPERUSERS are allowed to invoke the
  _COMMAND add-on.  If you want, you can require additional privileges
  (see the top of _COMMAND.CMD for details).