|
| 1 | +--- |
| 2 | +title: Developer's Guide |
| 3 | +layout: default |
| 4 | +parent: Community |
| 5 | +grand_parent: TrustGraph Documentation |
| 6 | +--- |
| 7 | + |
| 8 | +# Developer Guide |
| 9 | + |
| 10 | +This guide provides instructions for developers working on TrustGraph, covering build processes, release management, and local development. |
| 11 | + |
| 12 | +## Community contributors |
| 13 | + |
| 14 | +We welcome community contributions. We love constructive discussions and |
| 15 | +debates about AI, knowledge graphs, data sovereignty and responsible AI. |
| 16 | +Also, how to achieve these things through good engineering. |
| 17 | + |
| 18 | +### Philosophy |
| 19 | + |
| 20 | +TrustGraph values: |
| 21 | + |
| 22 | +- Radical Transparency and Open Source Accountability |
| 23 | + |
| 24 | + The community would prioritize complete transparency in AI |
| 25 | + operations, making all components inspectable, modifiable, and |
| 26 | + auditable. TrustGraph is "built with transparency and community |
| 27 | + collaboration in mind" and allows users to "easily inspect, modify, |
| 28 | + and extend the platform." This principle would extend beyond code to |
| 29 | + include decision-making processes, data handling, and algorithmic |
| 30 | + choices. |
| 31 | + |
| 32 | +- Knowledge Sovereignty and Reusability |
| 33 | + |
| 34 | + A core tenet would be empowering users to truly own and control |
| 35 | + their knowledge assets. TrustGraph enables "reusable Knowledge Cores |
| 36 | + that can be stored, shared, and reloaded," reflecting a philosophy |
| 37 | + that knowledge should be portable, persistent, and under user |
| 38 | + control rather than locked into proprietary systems. |
| 39 | + |
| 40 | +- Modular Interoperability Over Vendor Lock-in |
| 41 | + |
| 42 | + The community would champion architectural flexibility that prevents |
| 43 | + dependency on any single provider. TrustGraph provides "component |
| 44 | + flexibility" to "avoid component lock-in" and "integrates multiple |
| 45 | + options for all system components." This philosophy would extend to |
| 46 | + supporting diverse deployment environments and maintaining |
| 47 | + compatibility across different AI ecosystems. |
| 48 | + |
| 49 | +- Contextual Intelligence Through Structured Relationships |
| 50 | + |
| 51 | + Rather than treating data as isolated fragments, the community would |
| 52 | + emphasize understanding information through its connections and |
| 53 | + relationships. TrustRAG "understands and utilizes the relationships |
| 54 | + between pieces of information" and leverages "automatically |
| 55 | + constructed Knowledge Graphs to provide richer and more accurate |
| 56 | + context." This represents a philosophy that meaningful AI requires |
| 57 | + structured, interconnected understanding. |
| 58 | + |
| 59 | +- Collaborative Security and Distributed Trust |
| 60 | + |
| 61 | + The community would operate on principles of shared responsibility |
| 62 | + for AI safety and security, with trust built through collective |
| 63 | + verification rather than centralized authority. TrustGraph focuses |
| 64 | + on "providing a secure supply chain for AI components" while |
| 65 | + maintaining open-source accessibility, suggesting a model where |
| 66 | + security emerges from community oversight and collaborative |
| 67 | + validation. |
| 68 | + |
| 69 | +- Automate everything |
| 70 | + |
| 71 | + We have a lot of integrating to do. |
| 72 | + |
| 73 | +If it isn't obvious we value well-architected software. Good software |
| 74 | +unlocks all of the above values. Well-architected software prizes |
| 75 | +innovation as well as maturity. So, we value architectural principles |
| 76 | +which allow innovative software and experimental capability to sit alongside |
| 77 | +robustly mature software. |
| 78 | + |
| 79 | +### Use of git |
| 80 | + |
| 81 | +#### Branches |
| 82 | + |
| 83 | +- Release branches: `release/X.Y` e.g. `release/1.0` |
| 84 | +- Feature branches: `feature/FEATURE-NAME` e.g. `feature/authentication` |
| 85 | +- Maintenance branches: `maint/MAINT-NAME` e.g. `maint/update-pulsar-deps` |
| 86 | +- Bugfix branches: `fix/FIX-NAME` e.g. `fix/gateway-proto-failure` |
| 87 | + |
| 88 | +#### Tags |
| 89 | + |
| 90 | +- Release tags: `vX.Y.Z` e.g. `v1.2.3` |
| 91 | + |
| 92 | +## Release Management |
| 93 | + |
| 94 | +### Release Process |
| 95 | + |
| 96 | +1. **Prepare Git Repository** |
| 97 | + |
| 98 | + Ensure your repository is clean and ready for release. |
| 99 | + |
| 100 | +2. **Tag the Release** |
| 101 | + |
| 102 | + Create and push a version tag: |
| 103 | + |
| 104 | + ```bash |
| 105 | + git tag -a v1.2.3 -m '' |
| 106 | + git push --tags |
| 107 | + ``` |
| 108 | + |
| 109 | +3. CI/CD pipelines deliver the results to Docker Hub |
| 110 | + |
| 111 | +## Local Development |
| 112 | + |
| 113 | +### Building Locally |
| 114 | + |
| 115 | +You need a good configuration to work with. Testing with Podman Compose |
| 116 | +works best. To make a configuration, either: |
| 117 | +- be prepared to create a configuration and modify it: |
| 118 | + https://config-ui.demo.trustgraph.ai, or |
| 119 | +- Study how to build configurations with the underlying template tool: |
| 120 | + https://github.com/trustgraph-ai/trustgraph-templates |
| 121 | + |
| 122 | +### Building product |
| 123 | + |
| 124 | +The core system consists of many package types, but they have different |
| 125 | +value chains. The service deployment is a set of containers, built to |
| 126 | +include packages, so the target environment is a Docker Compose file |
| 127 | +launched to execute the containers. |
| 128 | + |
| 129 | +Whereas client-side, Python packages are enough to execute capability |
| 130 | +which interacts with running services. |
| 131 | + |
| 132 | +#### Building Python packages |
| 133 | + |
| 134 | +``` |
| 135 | +make packages VERSION=x.y.z |
| 136 | +``` |
| 137 | + |
| 138 | +This creates packages under the `dist` directory. You can then use `pip` |
| 139 | +to install them. |
| 140 | + |
| 141 | +#### Building containers |
| 142 | + |
| 143 | +``` |
| 144 | +make container VERSION=x.y.z |
| 145 | +``` |
| 146 | + |
| 147 | +You can then use a `docker-compose.yaml` file to run them. Such a file |
| 148 | +can be created from the TrustGraph config utility or the |
| 149 | +`trustgraph-templates` utility repo. |
| 150 | + |
0 commit comments