master control program
CMS MCS shell
The launch link leads to a MCP¹ shell provisioned from core domain space or the AWS, linode, or LAN/DIY inventory ² configured in the KEE Dashboard Control Block in your home profile. A MCS is a message control system operating under a MCP. ABORTED, ACTIVE, INUSE, COMPLETEDOK, or STOPPED are the possible job completion codes ³ for the launch attempt from the DS station at which you're currently authenticated.
Configuring a device control block for it in your profile dashboard enables system port management, or you can manually manage ssh as usual (e.g using ssh-copy-id).
Launch states ABORTED, COMPLETED, OR STOPPED, imply receipt of diagnostic info by the currently selected means of notification in your home profile (icon above, left of launch status). ABORTED is the immediate result if you are not authenticated.
A running SPO counts as a against account launch limits.
¹Core space resources are only dynamically allocated for AKPERSONs.
² AWS, Linode, or your Gb LAN. MCP nodes must have sub-millisecond ping. Set cloud vendor credentials in your commons dashboard or run the MCP leader in your LAN.
The namestyle is a homage to MCP and VM/CMS.
While initially only linux is supported, ultimately heterogeneous and in particular emulated mainframe kernel OS support is intended. A command and edit (CANDE) MCS wraps the ssh protocol.
Operators use a CANDE MCS for MCP command line ops which can use:
- and any of
mcpcms, the default, a custom zsh.
- shcl (common lisp nature)
- shhs (HsShellScript, haskell nature)
- upsh (prolog nature)
Only mcpcms need be set in /etc/shells, the others are available as the listed commands in a mcpcms session. The sources for the versions forked for MCP and how they operate there as opposed to their original authors intents are in my github repos. CANDE may be local or MCP or DCP depending on the operators context.
GHC is the haskell implementation. Lisp and prolog implementations are variable, and multiple can be combined but shcl and upsh themselves use sbcl and swipl, respectively.
The mcpcms level is always present, the others are outer shells adapted for the domain space knowledge engineering context.
The KEE uses the three HOLs² listed but the mcpcms level is appropriate for regular command line ops in MCP.
Although diverged for DS, an effort will be made to track changes in still vital original lang specific shells.
Assuming the target is configured and ready, invoke mcpcms with:mcpcms <req-spec> where <req-spec> ::= ipV6Address:port | ipV4Address:port | FQDSAgentName FQDSAgentName ::= <agentId>@<domain>[:<port>] and the port is displayed in a DS control block in the user's DCMS account profile or provided dynamically by automation in an authenticated session.
The port cannot be 22. The semantics are different depending on whether addr:port or agentId forms are used.
The bare address:port forms presume the user's client OS account name is a valid agentId and DCMS username. In this case the address is a pre allocated MCP element and ssh could be used instead.
The agentId form assigns existing or allocates a new MCP element from context implicit in the users account. In this case the CANDE MCS is used. node station.
- 1.0.0 PoC: working DCP provisioning of composed MCP images and mcpcms hosts via the DCMS> C-六 backend.
- 1.1.0 First working SPO and WFL development server.
- 1.2.0 Consolidation and productization release.
- 2.0.0 Mature DDD/KEE product.
² High Order Language ³ The states COMPLETED, COMPILEDOK and SCHEDULED are invalid for the launch WFL job, but its task steps may have COMPLETED codes, usable in other interfaces.