Robyn's plan for 2023! ✨

[sansyrox]

Hey All 👋

2022 was an amazing year for us and we hope that 2023 will be even better. I recently wrote an article telling about our plans for 2023, you can check it out here(https://sanskar.wtf/posts/robyns_plans_for_2023)

In this issue, I would like to talk about a more technical part

• implement request validation
  • here I wanted to implement some sort of message validation, and @kliwongan is doing an amazing job here https://github.com/sansyrox/robyn/pull/363

~- improve how you get body data~ ~ - when you access the response body, we still need to decode the byte array. I feel that this is an unnecessary step ~ ~ and we should be returning a saner response type by default.~

~- add some sort of performance benchmark integration~ ~ - I want to add a performance benchmark in our CI. This way we are certain if the project is becoming slower or faster~ ~in everychange. I have been given an early access to codspeed(https://codspeed.io/) and I would like to enable the~ ~python integration for it.~

• Fix the testing and deployment pipelines(remove the preview build and implement a better version of preview build by building on every merge on main)

  • Right now, I hate how we don't have a version to test before releasing it to the public PyPi. Plan for correction:
    • Delete the test PyPi action
    • Build a package on every merge to main, and be able to download it
    • Publish to PyPi on every release

• ~fix the default port~

  • ~Apple has unfortunately taken over the port 5000 . It is very annoying to change the port every time we test for~
  • ~the default settings. I plan to moving the default port to 8080 as 3000 is the default port for various front end~
  • ~frameworks.~

• openapi support

  • More or less self explanatory. I want to be able to have a route that automatically generates openapi spec for a web framework.

• add support for yields

  • I want to allow yields in the framework. That way we can do some processing after the response has been returned

• complete Strawberry GraphQl support

  • We have an initial integration with Strawberry but we need to create three things
  • Implement class views
  • Implement functional views
  • Add code to the strawberry-graphql repo

• ~allow dependency injection~ -~ I love how we can have dependency injection in FastAPI, I would love to add it in Robyn too.~

• then add a first class support for the database orm

  • The lack of database support is something that I don't enjoy. I want to add a first class orm support for a database but since Robyn is also a microframework, we would still want to allow the ability swap it out. Like our jinja templates.
  • After this, we can work on create an end to end template for robyn and frontend frameworks like asked for here(https://github.com/sansyrox/robyn/issues/360)

• improving the educational content around Robyn

  • A project is only as good as its documentation. Improve the documentation and video content around Robyn.

• ASGI ;)

• ~Improve support for windows~

  • ~The support for windows has been a bit lacklustre in the past. Improve support.~

• ~add cors support~

• ~-m flag should add some sort of documentation or helpers~

  • ~This is a good to have experiment. Can we have an ability to run Robyn as a module that will give us a link to~
  • ~thedocumentation or the entire documentation itself?~

• ~allow different mime types~

  • ~Also, inspect how static files are being handled here.~

• Improve websockets

  • Web sockets are not the fastest right now. Need to improve the speed.
  • allowing blocking options in async web sockets

• Implement SSE

• Default authorization and authentication

  • Add the ability to add default authorization and authentication. But like the templating support, this needs to hot swappable.

• route splitting across multiple files

  • Currently the only way to add routes in Robyn is through a single file. It is not scalable for a large codebase. We need to allow the ability to add multiple routes.

• Automatic const request optimization

  • https://github.com/sansyrox/robyn/issues/209

• middlewares:

  • see if the responses are handled the right way in middlewares(https://github.com/sansyrox/robyn/issues/244)
  • implement chainable middlewares(https://github.com/sansyrox/robyn/issues/206)

Somethings left to do from 2022:

• Make a community of plugins
• Improve docs website
• Better jsonify method

Feel free to comment with what you think of it and if there are any features that you'd like to have in this roadmap.

[XChikuX]

Better jsonify method

Please use a highly efficient json serializer and de-serializer like: Orjson If you are planning to do it in pure python.

---

Please keep pydantic data validation in mind wherever it makes sense.

[baseplate-admin]

Better jsonify method

Please use a highly efficient json serializer and de-serializer like: Orjson If you are planning to do it in pure python.

Please keep pydantic data validation in mind wherever it makes sense.

Why not use serde-json for even faster json parsing ?

[sansyrox]

@baseplate-admin , because the last time I checked serde-json required types in the rust side/compile side. However, we want to have an ability to serialize and deserialize without having compile-time typing.

Are you aware if it is possible to achieve this now?

[baseplate-admin]

Apologies. My knowledge in rust is modest at best.

My finds came from OrJSON. As OrJson uses serde_json

https://github.com/ijl/orjson/blob/c2514ab9ecdb9d42945f3fc6e18475b88bef8ec2/Cargo.toml#L59-L60

I was thinking to apply the logics at rust backend level to remove the python overhead.

[StandinKP]

@sansyrox Is this Roadmap updated for this year? What are we planning on adding this year?

[sansyrox]

Hey @StandinKP 👋

I have an updated Roadmap, will publish by next week :D