As for visible 3D concrete structure display, we need revit or blender to build a 3D model, and ifcjs to display it on the web.
As to knowledge display, revealjs should be use, so that the users could show their markdown through it easily.
As to backend,
To build a knowledge graph, neo4j in Java is a not a hyper choice, we are glad to try agdb as our graph database.
For symbolic computation, python's sympy is the first choice while the other's far away useful enough. In rust neither symbolic_polynomials nor savage is far from easily useful.
Run python on server and send back the result. Here are two ways: use socket or severless to deploy text knowledge and formula separately on server side; Or just use pyO3 to build our server.
Deploy
We could not use Vercel maybe, for our backend is almost in Rust. There are two choice, just using an Aliyun server or try serverless(FaaS) deployment. They are challengeable.
Implementation
The most difficult part is formula computation or symbolic math as we know. Maybe users can do something like zed's GPT assistant implementation: user select the formulas and select an operation, send it to backend and got the result.
This is quite difficult for we don't know how to translate markdown's formula like y = x^2 to sympy's expr y = x ** 2. The best solution I can post is to use ChatGPT's api for now.
Everyone's jobs && Language
We should also divide the task apart, so any of us can got a job.
Apart from that, I'd like to make this project more well-known, so the language the first contributor using is quite important. I prefer English, however, the knowledge on our first demo, and the users we may have are all Chinese, so Chinese is also OK.
We should finish
rfc0001 Tech-Selection
, although none will comment this now maybe.However, it's quite important to select the tech we'll use. As for now, I suggest using
Rust
,python
andtypescript
as the language we use.To build a web app:
As to frontend
nextjs
andreact
are both important.revit
orblender
to build a 3D model, andifcjs
to display it on the web.As to backend,
sympy
is the first choice while the other's far away useful enough. In rust neither symbolic_polynomials nor savage is far from easily useful.Deploy
We could not use Vercel maybe, for our backend is almost in Rust. There are two choice, just using an
Aliyun
server or try serverless(FaaS
) deployment. They are challengeable.Implementation
The most difficult part is formula computation or symbolic math as we know. Maybe users can do something like
zed
's GPT assistant implementation: user select the formulas and select an operation, send it to backend and got the result.This is quite difficult for we don't know how to translate markdown's formula like
y = x^2
to sympy's expry = x ** 2
. The best solution I can post is to use ChatGPT's api for now.Everyone's jobs && Language
We should also divide the task apart, so any of us can got a job.
Apart from that, I'd like to make this project more well-known, so the language the first contributor using is quite important. I prefer English, however, the knowledge on our first demo, and the users we may have are all Chinese, so Chinese is also OK.