# Solution Design: msghandler

**Version**: 1.3.0  
**Date**: 2026-05-22  
**Status**: Active  
**Ground Truth**: [`src/msghandler.jl`](../src/msghandler.jl)

---

## 1. Problem Decomposition

msghandler addresses the challenge of cross-platform data exchange between **Julia**, **JavaScript**, **Python**, **Dart**, **Rust**, and **MicroPython** applications using message brokers as transport layers.

### Problem Statement

Developers working across multiple programming languages face significant obstacles when trying to share data:

| Problem | Description | User Impact |
|---------|-------------|-------------|
| **P-001**: Cross-platform data serialization | Different languages have incompatible data types and serialization formats | Developers must write platform-specific conversion code |
| **P-002**: Large payload handling | Message brokers have size limits, but large files need to be transferred | Large files either fail or require complex workarounds |
| **P-003**: Transport abstraction | Each platform has different message broker libraries and APIs | No unified interface across platforms |
| **P-004**: Request-response patterns | Bi-directional communication requires complex correlation tracking | Developers must implement custom message routing |

### Solution Boundaries

**In Scope**:
- Unified API for `smartpack()` and `smartunpack()` across all platforms
- Automatic transport selection based on payload size
- File server integration using Claim-Check pattern
- Multi-payload support with mixed types in single message
- Exponential backoff for reliable file downloads

**Out of Scope**:
- Message compression (adds complexity without clear benefit)
- Message encryption (application-layer concern)
- Advanced message routing (simple topic matching sufficient)
- Persistent message queues (transport pattern sufficient)

### Decision IDs

| Decision ID | Decision | Description |
|-------------|----------|-------------|
| SD-001 | Claim-Check Pattern | Large payloads uploaded to HTTP server, small payloads sent directly |
| SD-002 | Automatic Transport Selection | <0.5MB = direct, ≥0.5MB = link based on size threshold |
| SD-003 | Handler Function Abstraction | Pluggable file server implementations via handler functions |
| SD-004 | Unified Tuple Format | Same `(dataname, data, type)` format across all platforms |
| SD-005 | Base64 Encoding | JSON-compatible binary data transport |
| SD-006 | Transport Abstraction | Support multiple broker protocols (NATS/MQTT/WebSocket) transparently |

---

## 2. Solution Approach

msghandler implements a **Claim-Check pattern** with intelligent transport selection:

```
Sender (smartpack)              Transport Layer              Receiver (smartunpack)
┌─────────────────┐             ┌───────────────┐            ┌───────────────────┐
│                 │             │               │            │                   │
│ 1. Data tuples  │────────────>│               │───────────>│ 1. Parse envelope │
│    [(name,      │   JSON      │  Message      │   JSON     │ 2. Check transport│
│     data, type)]│   format    │  Broker       │   format   │ 3. Fetch/Decode   │
│                 │             │  (NATS/MQTT/  │            │ 4. Return tuples  │
└─────────────────┘             │  WebSocket)   │            │                   │
                                │               │            └───────────────────┘
                                └───────────────┘
```

### Key Design Decisions

| Decision ID | Decision | Rationale | Alternatives Rejected |
|-------------|----------|-----------|----------------------|
| **SD-001** | Claim-Check Pattern | Large payloads (>0.5MB) uploaded to HTTP server, small payloads sent directly via transport | Client-side compression - adds complexity; Server-side compression - not universally supported |
| **SD-002** | Automatic Transport Selection | <0.5MB = direct (fast), ≥0.5MB = link (avoid transport limits) | Manual selection - error-prone; Fixed threshold - not adaptive |
| **SD-003** | Handler Function Abstraction | Allows pluggable file server implementations (Plik, AWS S3, custom) | Hardcoded Plik - not flexible; Interface-based - too complex for this use case |
| **SD-004** | Unified Tuple Format | Same input/output format across all platforms | Platform-native formats - no interoperability; Protocol buffers - too heavy |
| **SD-005** | Base64 Encoding | JSON-compatible binary data transport | Raw bytes - not JSON-compatible; Hex encoding - 2x size overhead |
| **SD-006** | Transport Abstraction | Support multiple broker protocols (NATS/MQTT/WebSocket) transparently | Platform-specific libraries - no interoperability |

### Architecture Components

```mermaid
flowchart TB
    subgraph Client["Client Application"]
        direction TB
        APP["Application Code"]
        API["msghandler API"]
        
        APP -->|Data tuples| API
        API -->|JSON envelope| TRANSPORT
    end
    
    subgraph Transport["Transport Layer"]
        direction TB
        BROKER["Message Broker<br/>NATS/MQTT/WebSocket"]
        TOPICS["Topic Subscription"]
        
        API -->|Publish| BROKER
        BROKER -->|Deliver| TOPICS
        TOPICS -->|Subscribe| API
    end
    
    subgraph FileServer["File Server"]
        direction TB
        UPLOAD["Upload Handler"]
        DOWNLOAD["Download Handler"]
        
        API -.->|Upload URL| UPLOAD
        DOWNLOAD -.->|Fetch URL| API
    end
    
    style CLIENT fill:#e1f5fe,stroke:#0288d1,stroke-width:2px
    style Transport fill:#ffe0b2,stroke:#f57c00,stroke-width:2px
    style FileServer fill:#c8e6c9,stroke:#43a047,stroke-width:2px
```

---

## 3. Alternatives Considered

| Alternative | Pros | Cons | Decision |
|-------------|------|------|----------|
| **gRPC/Protobuf** | Strong typing, efficient binary format | No native MicroPython support; Complex schema management | Rejected - not cross-platform enough |
| **MessagePack** | Compact binary, good performance | Browser support limited; No standard for tabular data | Rejected - missing Arrow IPC alternative |
| **Protocol Buffers** | Type-safe, efficient | No native support for tabular data exchange | Rejected - cannot represent DataFrames natively |
| **REST HTTP Upload** | Simple, universal | High latency; No real-time capability | Rejected - not suitable for message broker pattern |
| **Hybrid (direct/link)** | Optimal for both small and large payloads | More complex implementation | Accepted - matches user requirements (FR-003, FR-004) |
| **Single transport type** | Simpler implementation | Cannot handle large payloads efficiently | Rejected - violates FR-003 requirement |
| **Platform-specific APIs** | Native performance | No interoperability; Maintenance burden | Rejected - violates cross-platform goal |

---

## 4. High-Level Component Diagram

```mermaid
flowchart TD
    subgraph msghandler["msghandler Core Module"]
        direction TB
        
        subgraph Serialization["Serialization Layer"]
            DIR["Direct Transport"]
            LNK["Link Transport"]
            
            DIR -->|Base64| JSON_MSG
            LNK -->|HTTP URL| JSON_MSG
        end
        
        subgraph Envelope["Envelope Builder"]
            HDR["Message Header"]
            PAY["Payload Manager"]
            
            HDR --> PAY
        end
        
        subgraph Handlers["Handler Functions"]
            UPD["Upload Handler"]
            DWN["Download Handler"]
            
            UPD --> LNK
            DWN --> LNK
        end
        
        API["smartpack() / smartunpack()"]
        
        API -->|Input| Serialization
        API -->|Output| Serialization
        API -->|Configure| Handlers
    end
    
    subgraph Transport["Transport Layer"]
        BROKER["NATS / MQTT / WebSocket"]
        API -->|JSON| BROKER
        BROKER -->|JSON| API
    end
    
    subgraph FileServer["File Server"]
        Plik["HTTP Server"]
        UPD -.->|POST| Plik
        Plik -.->|URL| DWN
    end
    
    style msghandler fill:#b3e5fc,stroke:#0288d1,stroke-width:2px
    style Transport fill:#ffe0b2,stroke:#f57c00,stroke-width:2px
    style FileServer fill:#c8e6c9,stroke:#43a047,stroke-width:2px
```

### Component Responsibilities

| Component | Responsibilities | Decision IDs | Requirements Addressed |
|-----------|-----------------|--------------|----------------------|
| **Serialization Layer** | Convert data types to transport format (Base64/URL) | SD-005 | FR-001, FR-002, FR-012 |
| **Envelope Builder** | Create standardized message envelope with metadata | SD-001 | FR-011, FR-013, FR-014 |
| **Handler Functions** | Abstract file server operations for pluggability | SD-003 | FR-008, FR-009 |
| **Transport Adapter** | Support multiple broker protocols transparently | SD-006 | FR-013, FR-014 |
| **Payload Manager** | Track payload types, sizes, and encoding | SD-004 | FR-006, FR-007 |

---

## 5. Decision Rationale

### SD-001: Why Claim-Check Pattern?

**Requirement**: FR-003 - Large file handling, FR-004 - Direct transport for small payloads

**Rationale**:
- Transport layers (NATS, MQTT) have message size limits (typically 1MB)
- Direct transport is faster for small payloads (no file server round-trip)
- Link transport avoids transport limits for large payloads
- User doesn't need to manually choose - automatic selection based on threshold

### SD-002: Why Handler Functions for File Server?

**Requirement**: FR-008 - Plik integration, FR-009 - Custom file server support

**Rationale**:
- Plik is common open-source solution for file server
- Some users need AWS S3 or custom implementation
- Handler functions provide clean abstraction without vendor lock-in
- Same signature across all platforms (unified API)

### SD-003: Why Tuple Format for Payloads?

**Requirement**: FR-006 - Multi-payload messages, FR-007 - Payload type preservation

**Rationale**:
- `(dataname, data, type)` tuple is language-agnostic
- Simple to understand: name, content, type
- Supports mixed payload types in single message
- Easy to serialize/deserialize across platforms

### SD-004: Why Base64 Encoding?

**Requirement**: FR-012 - Message serialization, FR-001 - Cross-platform text messaging

**Rationale**:
- JSON is universal - works on all platforms
- Base64 converts binary to ASCII for JSON compatibility
- Standard format with native support in all languages
- No additional dependencies needed

### SD-005: Why Automatic Transport Selection?

**Requirement**: FR-003, FR-004, NFR-104, NFR-105

**Rationale**:
- <0.5MB payloads use direct transport (<1s latency, FR-004 KPI)
- ≥0.5MB payloads use link transport to avoid transport limits (FR-003 KPI: 99% successful uploads)
- User doesn't need to manually choose - automatic selection based on threshold

### SD-006: Why Transport Abstraction?

**Requirement**: FR-013, FR-014, NFR-201

**Rationale**:
- Support multiple broker protocols (NATS, MQTT, WebSocket) transparently
- Caller handles actual transport publishing/subscription
- Unified API across all platforms
- At-least-once delivery semantics via transport layer

---

## 6. Risk Assessment

| Risk | Impact | Probability | Mitigation |
|------|--------|-------------|------------|
| **Performance degradation with >500KB payloads** | High | Medium | Size threshold detection; Link transport fallback |
| **File server availability issues** | Medium | Low | Exponential backoff retry; Graceful degradation |
| **Platform-specific bugs** | Medium | Low | Comprehensive test suite per platform; CI validation |
| **Encoding mismatches between platforms** | High | Low | Strict specification; Test contracts; Validation rules |
| **Transport layer incompatibility** | Medium | Low | Transport-agnostic design; Handler abstraction |

---

## 7. Requirements Traceability

| Solution Component | Decision ID | Requirement ID | Description |
|-------------------|-------------|----------------|-------------|
| **smartpack() function** | SD-001, SD-002, SD-004, SD-005, SD-006 | FR-001, FR-002, FR-003, FR-004, FR-005, FR-006, FR-007, FR-008, FR-009, FR-010, FR-011, FR-012, FR-013, FR-014 | Unified API for sending messages across all platforms |
| **smartunpack() function** | SD-001, SD-002, SD-004, SD-005, SD-006 | FR-001, FR-002, FR-003, FR-004, FR-005, FR-006, FR-007, FR-008, FR-009, FR-010, FR-011, FR-012, FR-013, FR-014 | Unified API for receiving messages across all platforms |
| **Direct transport** | SD-002 | FR-004, NFR-101, NFR-102, NFR-104, NFR-105 | Send payloads < threshold directly via transport |
| **Link transport** | SD-001, SD-002 | FR-003, NFR-104, NFR-105 | Upload payloads ≥ threshold to file server |
| **File server handler** | SD-003 | FR-008, FR-009, FR-010 | Pluggable upload/download handlers with retry logic |
| **Payload type preservation** | SD-004 | FR-006, FR-007 | Support text, dictionary, arrowtable, jsontable, image, audio, video, binary |
| **Correlation ID** | SD-001 | FR-011, NFR-401, NFR-403 | Message tracing across distributed systems |
| **Multi-payload support** | SD-004 | FR-006, FR-007 | List of (dataname, data, type) tuples |

### Non-Functional Requirements Traceability

| Solution Component | Decision ID | NFR ID | Description |
|-------------------|-------------|--------|-------------|
| **Serialization optimization** | SD-005 | NFR-101, NFR-102 | <50ms overhead for 10KB payloads |
| **Transport efficiency** | SD-006 | NFR-103 | <100ms connection establishment |
| **File server latency** | SD-001, SD-002 | NFR-104, NFR-105 | <1s upload/download for 0.5MB files |
| **Concurrent connections** | SD-006 | NFR-106 | Support 100+ simultaneous connections |
| **Message throughput** | SD-005, SD-006 | NFR-107 | Handle 1000+ messages/second per instance |
| **At-least-once delivery** | SD-006 | NFR-201 | Transport layer semantics |
| **Graceful degradation** | SD-003 | NFR-202 | File server unavailability handling |
| **Auto-reconnect** | SD-006 | NFR-203 | Transport connection failure recovery |
| **Required logs** | SD-001 | NFR-401 | Correlation ID, msg_id, timestamp, etc. |
| **Critical metrics** | SD-001, SD-005 | NFR-402 | messages_sent_total, file upload/download duration |
| **Tracing** | SD-001 | NFR-403 | Correlation ID propagation |

---

## 8. Gap-Check Validation

| Stage Transition | Gap-Check Question | Status |
|------------------|-------------------|--------|
| **Requirements → Solution Design** | Does the Solution Design clearly explain how the system solves the user problem, not just what it does? | ✅ Verified - All user stories mapped to solution components with requirement ID and decision ID references |
| **Solution Design → Specification** | Does the Specification define all technical details that the solution approach requires? | ⏳ Pending - Specification needs review for completeness |
| **Solution Design → Walkthrough** | Does the Walkthrough reflect the complete flow including error states and timing? | ⏳ Pending - Walkthrough needs validation against design |

### Solution Design Validation

**Problem**: Users need to send mixed payload types (text + image + large file) between Julia, JavaScript, Python, and MicroPython applications.

**Solution Components**:
1. **SD-001** - `smartpack()` - Unified API for all platforms
2. **SD-002** - Tuple format - `(dataname, data, type)` - platform-agnostic
3. **SD-003** - Automatic transport selection - <0.5MB = direct, ≥0.5MB = link
4. **SD-004** - File server handler abstraction - Plik/AWS S3/custom support
5. **SD-005** - Exponential backoff - Reliable file downloads
6. **SD-006** - Correlation ID - Message tracing

**Requirement Mapping**:
- FR-001, FR-002, FR-003, FR-004, FR-005, FR-006, FR-007, FR-008, FR-009, FR-010, FR-011, FR-012, FR-013, FR-014 ✅

**Gap Check**: Does this solution explain *how* users will actually use the system?

**Answer**: Yes - the walkthrough provides concrete examples:
1. JavaScript sends `[(msg, "Hello", "text"), (avatar, binary_data, "image")]`
2. `smartpack()` automatically selects transport based on size (SD-002)
3. Large file (≥0.5MB) → link transport → file server upload (SD-001)
4. Small payload (<0.5MB) → direct transport → base64 encoding (SD-005)
5. Receiver calls `smartunpack()` → receives same tuple format

---

*This solution design document is versioned and maintained in git alongside the codebase. All implementations must adhere to this design.*

**Traceability Summary**:
- All requirements traced to solution components with SD-XXX decision IDs
- Each decision ID references the corresponding requirement IDs (FR-XXX, NFR-XXX)
- Specification must cite SD-XXX references for each technical detail