What to expect in 2025? Teo 0.4.0 planning

In the year end of 2024, I published Teo 0.3.8, which is the best version for now. The features are complete. The supported platforms are complete, too.

A backend framework should have great user experience. I look at the current feature set of Teo, and there are still things need to be improved. Current issues include:

1. The language parser and type checker is slow
2. The hooking point of writing custom handlers and pipe segments are not intuitive.
3. For some developers, generating the entities are annoying. They tend to prefer writing in a single language.
4. The ORM apis are not as great as synthesized handlers and they cause complexities.
5. The permission checking is not easy to use.

I designed a new version with breaking changes to fix the all the above issues and bring great new features.

Language

The new Teo language is a great overhaul over the current one. Scripting ability is added. With this, we could write permission checking easier. And even more, we could write route handlers, pipe segments in the scripting files.

The new language unifies handler, decorators, and pipe segments into function. The name conflicts if duplicated. This is for the sake of extending Teo with programming language traits and classes.

Language parser and type checker

The performance of the current parser and type checker is noticeably slow. To support scripting and increase performance. I decide to rewrite the parser and type checker. I will also introduce a interpreter for developers to play around TEO language.

Removal of ORM api

The ORM object can persist even out of transactions. This created a lot of complexities when I implemented it. The ORM api lacks some features like upsert. I decide to remove ORM api and introduce a better way to interact the database which is fragment.

Fragment

Fragment is a piece of code, which is an async function that takes a TEON input and returns a TEON output. Let's say, to create a database record, we take the data input, selected output fields and selected relations, to manipulate the database, and generate the output. Current synthesized handlers all do this way. Fragment should be used internally and it's the building block of user's HTTP API set.

Fragment can do permission checking by @canAccess. The synthesized handlers will just be wrappers around the synthesized fragments.

User can define his own fragments and call the fragments in the fragments. This way, we unified the API code sharing and database access. A fragment named sql will be synthesized for SQL database environment, too.

No schema

The new version of Teo supports pure programming language. The schema is not required to run TEO servers. Define configs, declare models and handlers, directly with programming language like Node.js and Python.

ORM only & server only

Teo has a great ORM inside, I will update the API structure to allow developers to use the ORM or the server exclusively.

Release plan

This new version is planned to be released in May, 2025. And it may delay. Before the end of 2025 comes, it will be released anyway.