vise: A Constrained Size Output Virtual Machine

An attempt at defining a small VM to handle menu interaction for size-constrained clients and servers.

Rationale

Original motivation was to create a simple templating renderer for USSD clients, combined with an agnostic data-retrieval reference that may conceal any level of complexity.

Features

Implemented

• Define and enforce max output size for every individual output.
• Allow one single data entry to fill remaining available size capacity.
• An assembly-like mini-language to define:
  • external code execution.
  • input validation and routing.
  • menu definitions.
  • flow control.
  • exception handling.
• templated output from results of external code execution.
• generate and navigate pages where data symbol contents are too long to display on a single page.
• pluggable function design for handling external code execution calls.

Pending

• Node Walking Audit Tool (NWAT) to ensure all nodes produce output within constraints.
• Input generator engine for the NWAT.
• State error flag debugger representation, builtin as well as user-defined.
• Stepwise debug view on log/stderr of state mutations.
• Toolset to assist bootstrapping/recovering (failed) state from spec.

Possibly useful

• Breakpoints.
• Key/value database reference example.

Opcodes

The VM defines the following opcode symbols, alphabetically:

• CATCH <symbol> <signal> - Jump to symbol if signal is set (see signals below). If match, has same side-effect as move.
• CROAK <signal> - Clear state and restart execution from top if signal is set (see signals below). If match, has same side-effect as move.
• HALT - Stop execution. The remaining bytecode (typically, the routing code for the node) is returned to the invoking function.
• INCMP <arg> <symbol> - Compare registered input to arg. If match, it has the same side-effects as MOVE. In addition, any consecutive INCMP matches will be ignored until HALT is called.
• LOAD <symbol> <size> - Execute the code symbol symbol and cache the data, constrained to the given size. Can be exposed with MAP within scope. See "External code" below.
• MAP <symbol> - Expose a code symbol previously loaded by LOAD to the rendering client. Roughly corresponds to the global directive in Python.
• MNEXT <choice> <display> - Define how to display the choice for advancing when browsing menu.
• MOUT <choice> <display> - Add menu display entry. Each entry should have a matching INCMP whose arg matches choice. display is a descriptive text of the menu item.
• MOVE <symbol> - Create a new execution frame, invalidating all previous MAP calls.
• MPREV <choice> <display> - Define how to display the choice for returning when browsing menu.
• MSEP - Not yet implemented. Marker for menu page separation. Incompatible with browseable nodes.
• MSIZE <max> <min> - Not yet implemented. Set min and max display size of menu part to num bytes.
• RELOAD <symbol> - Execute a code symbol already loaded by LOAD and cache the data, constrained to the previously given size for the same symbol. See "External code" below.

External code

LOAD is used to execute code symbols in the host environment. It is loaded with a size constraint, and returned data violating this constraint should generate an error.

Any symbol successfully loaded with LOAD will be associated with the call stack frame it is loaded. The symbol will be available in the same frame and frames below it. Once the frame goes out of scope (e.g. BACK is called in that frame) the symbols should be freed as soon as possible. At this point they are not available to the abandoned scope.

Loaded symbols are not automatically exposed to the rendering client. To expose symbols ot the rendering client the MAP opcode must be used.

The associated content of loaded symbols may be refreshed using the RELOAD opcode. RELOAD only works within the same constraints as MAP. However, updated content must be available even if a MAP precedes a RELOAD within the same frame.

Methods handling LOAD symbols have the client input available to them.

External symbol optimizations

Only LOAD and RELOAD should trigger external code side-effects.

In an effort to prevent leaks from unnecessary external code executions, the following constraints are assumed:

• An explicit MAP must exist in the scope of any LOAD.
• All symbols declared in MAP must be used for all template renderings of a specific node.

Any code compiler or checked should generate an error on any orphaned LOAD or MAP symbols as described above.

Signals

Signal may be set when executing of external code symbols, and may be used as a simple exception mechanism.

The signal flag arguments should only set a single flag to be tested. If more than one flag is set, the first flag matched will be used as the trigger.

First 8 flags are reserved and used for internal VM operations.

When a signal is caught, the bytecode buffer is flushed before the target symbol code is loaded.

Avoid duplicate menu items

The vm execution should overwrite duplicate MOUT directives with the last definition between HALT instructions.

The assembler should detect duplicate INCMP and MOUT (or menu batch code) selectors, and fail to compile. MSEP should be included in duplication detection.

Menus

A menu has both a display and a input processing part. They are on either side of a HALT instruction.

To assist with menu creation, a few batch operation symbols have been made available for use with the assembly language.

• DOWN <symbol> <choice> <display> descend to next frame and move to symbol
• UP <choice> <display> return to the previous frame
• NEXT <choice> <display> include pagination advance
• PREVIOUS <choice> <display> include pagination return. If NEXT has not been defined this will not be rendered.

Rendering

The fixed-size output is generated using a templating language, and a combination of one or more max size properties, and an optional sink property that will attempt to consume all remaining capacity of the rendered template.

In this example

• maxOutputSize is 256 bytes long.
• template is 120 bytes long.
• param one has max size 10 but uses 5.
• param two has max size 20 but uses 12.
• param three is a sink.
• rendered menu is 15 bytes long.

The renderer may use up to 256 - 120 - 5 - 12 - 15 = 104 bytes from the sink when rendering the output.

Menu browsing

A max size can be set for the menu, which will count towards the space available for the template sink.

Menus too long for a single screen should be browseable through separate screens. How the browse choice is displayed is defined using the MSEP definition. The browse choice counts towards the menu size capacity.

When browsing additional menu pages, the template output should not be included.

Menu defaults

Browsing menu display definitions (MNEXT, MPREV) as well as size constaints (MSIZE) should have sane defaults defined by the assembler if they are missing from the assembly code.

Multipage support

Multipage outputs, like listings, are handled using the sink output constraints:

• first calculate what the rendered display size is when all symbol results that are not sinks are resolved.
• split and cache the list data within its semantic context, given the sink limitation after rendering.
• provide a next and previous menu item to browse the prepared pagination of the list data.

Languages support

Not yet implemeennted

Language for rendering is determined at the top-level state.

Lookups dependent on language are prefixed by either ISO 639-1 or ISO 639-3 language codes, followed by :.

Default language means records returned without prefix if no language is set. Default language should be settable at the top-level.

Node names must be defined using 7-bit ASCII.

Virtual machine interface layout

This is the version 0 of the VM. That translates to highly experimental.

Currently the following rules apply for encoding in version 0:

• A code instruction is a big-endian 2-byte value. See vm/opcodes.go for valid opcode values.
• symbol value is encoded as one byte of string length, after which the byte-value of the string follows.
• size value is encoded as one byte of numeric length, after which the big-endian byte-value of the integer follows.
• signal value is encoded as one byte of byte length, after which a byte-array representing the defined signal follows.

Reference implementation

This repository provides a golang reference implementation for the vise concept.

Structure

• asm: Assembly parser and compiler.
• cache: Holds and manages all loaded content.
• engine: Outermost interface. Orchestrates execution of bytecode against input.
• persist: Interface and reference implementation of state and cache persistence across asynchronous vm executions.
• render: Renders menu and templates, and enforces output size constraints.
• resource: Retrieves data and bytecode from external symbols, and retrieves templates.
• state: Holds the bytecode buffer, error states and navigation states.
• vm: Defines instructions, and applies transformations according to the instructions.

Template rendering

Template rendering is done using the text/template faciilty in the golang standard library.

It expects all replacement symbols to be available at time of rendering, and has no tolerance for missing ones.

Runtime engine

The runtime engine:

• Validates client input
• Runs VM with client input
• Renders result
• Restarts execution from top if the vm has nothing more to do.

There are two flavors of the engine:

• engine.Loop - class used for continuous, in-memory interaction with the vm (e.g. terminal).
• engine.RunPersisted - method which combines single vm executions with persisted state (e.g. http).

Client identification

The engine.Config struct defines a property SessionId which is added to the context.Context passed through entire engine vm call roundtrip.

This is used to identify the caller, and thus defines a top-level storage key under which data entries should be retrieved.

Bytecode examples

(Minimal, WIP)

000a 03666f6f 06746f20666f6f  # MOUT "foo" "to foo" - display a menu entry for choice "foo", described by "to foo"
0008 03666f6f 03626172        # INCMP "foo" "bar"   - move to node "bar" if input is "FOO"
0001 0461696565 01 01         # CATCH "aiee" 1 1    - move to node "aiee" (and immediately halt) if input match flag (1) is not set (1)
0003 04616263 020104          # LOAD "abc" 260      - execute code symbol "abc" with a result size limit of 260 (2 byte BE integer, 0x0104)
0003 04646566 00              # LOAD "def" 0        - execute code symbol "abc" with no size limit (sink)
0005 04616263                 # MAP "abc"           - make "abc" available for renderer
0007                          # HALT                - stop execution (require new input to continue)
0006 03313233                 # MOVE "123"          - move to node "123" (regardless of input)
0007                          # HALT                - stop execution

Assembly examples

See testdata/*.vis

Development tools

Located in the dev/ directory.

Test data generation

go run ./dev/gendata/ <directory>

Outputs bytecodes and templates for test data scenarios used in engine unit tests.

Interactive runner

go run ./dev/interactive [-d <data_directory>] [--root <root_symbol>] [--session-id <session_id>]

Creates a new interactive session using engine.DefaultEngine, starting execution at symbol root_symbol

data_directory points to a directory where templates and bytecode is to be found (in the same format as generated by dev/testdata).

If data_directory is not set, current directory will be used.

if root_symbol is not set, the symbol root will be used.

if session_id is set, mutable data will be stored and retrieved keyed by the given identifer (if implemented).

Assembler

go run ./dev/asm <assembly_file>

Will output bytecode on STDOUT generated from a valid assembly file.

Disassembler

go run ./dev/disasm/ <binary_file>

Will list all the instructions on STDOUT from a valid binary file.

Interactive case examples

Found in examples/.

Be sure to make examples before running them.

Can be run with e.g. go run ./examples/<case> [...]

The available options are the same as for the dev/interactive tool.

Contents of the case directory:

• *.vis - assembly code.
• *.bin - bytecode for each node symbol (only available after make).
• *.txt.orig - default contents of a single data entry.
• *.txt - current contents of a single data entry (only available after make).