Code Generation Tool#718
Conversation
✅ Deploy Preview for thingweb-playground ready!Built without sensitive environment variables
To edit notification comments on pull requests, go to your Netlify project configuration. |
egekorkan
left a comment
There was a problem hiding this comment.
Initial review based on git diff
- Netlify previews are failing due to dependencies. Not sure how the CI tests are passing if the problem is so basic :D
- For the tests, did you also do tests with a live Thing? like your coffee machine example or http://plugfest.thingweb.io/http-data-schema-thing ?
869d86c to
39bae0d
Compare
egekorkan
left a comment
There was a problem hiding this comment.
- Web package is working nicely.
- CLI:
- tsx as a dependency is missing
- Modbus and javascript is not working? Or is the base not used to detect protocol?
- When the output prompt step comes and I put a path,
~/Downloadsdoes not work
However, given that the CLI will live in td-tools, we can still go ahead and fix the issues in the commits to that repo
be40612 to
66470ea
Compare
Could you please share your inputs? Modbus serial works fine for me. |
The path input was only relative. I added support for absolute path inputs and an interactive mode for selecting the file or folder to ease the input. Could you test that again please? |
This is working. Modbus in Javascript with the following TD does not work. It is generated from https://github.com/wot-oss/thingmodels/blob/main/siemens/siemens/3nacom-fuse/v1.0.0-20240802121832-3b18ae135fbc.tm.json but use the attachment. The generation results in a prompt even though we have modbus binding |

This is a take over of PR: eclipse-thingweb/td-tools#61
The goal is to create a package of a code generation tool that creates a protocol specific code snippet in a chosen library of a programming language and integrate it to the playground.
Taking over the previous PR I could not make it run and had a different vision of the use of AI generation:
The Bring Your Own Token principal requires the playground of either:
In both cases the user needs a developer account in their respective LLM provider and has to know how to generate the token which reduces UX imo. Therefore, after discussing this with @egekorkan I implemented another concept where the tool generates a prompt that user can paste in an LLM of their own choice:
The tool is inside of this repository for now for testing purposes. Later it can be moved to the
td-toolsrepository.