5 ms·

I wish they never abandoned the goal of being a true Python superset. That would have been something.

JBits 2mo ago
I don't think the goal of being a true Python superset ever existed. Given that there are so many failed attempts to optimise arbitrary Python code, it was dubious from the start. Furthermore, the documentation completely ignored these challenges and made it clear that Mojo's goal was to compile to MLIR in order to write kernels for machine learning. The first example of Mojo code was a low level matrix multiplication kernel taking advantage of fancy compilation features and the documentation advertised pointers inspired by Rust's move semantics, neither of which are typically considered friendly or relevant to high level programmers.
- giancarlostoro 2mo ago
  It was in their roadmap: https://forum.modular.com/t/mojo-as-a-python-superset/2490 https://forum.modular.com/t/mojo-as-a-python-superset/2490
  - JBits 2mo ago
    I think it was there for marketing purposes and that there was little actual intention to work on it. From the roadmap: >Mojo may or may not evolve into a full superset of Python, and it's okay if it doesn't.
    - giancarlostoro 2mo ago
      That's a bold claim, I would be surprised if that were why they made the decision, there's other Python-like languages that don't need to pretend to be a superset to be successful.
      - JBits 2mo ago
        Being a superset of Python may not be necessary to be successful, but perhaps non-committally proposing it was necessary for Modular to win the funding they wanted. While lots of people indirectly run GPU kernels with PyTorch, I think the number of people wanting to write their own kernels is considerably smaller and that this niche already has established players (plenty of whom even already build on top of MLIR), making Mojo's most obvious use case difficult to sell.
- mastermage 2mo ago
  it definetely existed and was one of the major things why I was interested in it. I wished I could have used this to slowly migrate away from Python codes in Physics. But alas they dropped the goal and for me the ball with the decision to stop following this. I will probably not write mojo code when I could use Rust or Julia instead, which are more aligned with my necessities.
Mond_ 2mo ago
I don't think this goal was ever realistic, it was just a marketing pitch to get a foot in the door.
dlahoda 2mo ago
It still feels python enough.
melodyogonna 2mo ago
It is still a goal. Mojo's semantics are still driven by Pythons, and can only deviate where safety or performance is an issue. Eventually it'll get the dynamic parts of Python

JBits 2mo ago

I don't think the goal of being a true Python superset ever existed. Given that there are so many failed attempts to optimise arbitrary Python code, it was dubious from the start. Furthermore, the documentation completely ignored these challenges and made it clear that Mojo's goal was to compile to MLIR in order to write kernels for machine learning. The first example of Mojo code was a low level matrix multiplication kernel taking advantage of fancy compilation features and the documentation advertised pointers inspired by Rust's move semantics, neither of which are typically considered friendly or relevant to high level programmers.

giancarlostoro 2mo ago
It was in their roadmap: https://forum.modular.com/t/mojo-as-a-python-superset/2490 https://forum.modular.com/t/mojo-as-a-python-superset/2490
- JBits 2mo ago
  I think it was there for marketing purposes and that there was little actual intention to work on it. From the roadmap: >Mojo may or may not evolve into a full superset of Python, and it's okay if it doesn't.
  - giancarlostoro 2mo ago
    That's a bold claim, I would be surprised if that were why they made the decision, there's other Python-like languages that don't need to pretend to be a superset to be successful.
    - JBits 2mo ago
      Being a superset of Python may not be necessary to be successful, but perhaps non-committally proposing it was necessary for Modular to win the funding they wanted. While lots of people indirectly run GPU kernels with PyTorch, I think the number of people wanting to write their own kernels is considerably smaller and that this niche already has established players (plenty of whom even already build on top of MLIR), making Mojo's most obvious use case difficult to sell.
mastermage 2mo ago
it definetely existed and was one of the major things why I was interested in it. I wished I could have used this to slowly migrate away from Python codes in Physics. But alas they dropped the goal and for me the ball with the decision to stop following this. I will probably not write mojo code when I could use Rust or Julia instead, which are more aligned with my necessities.

Mond_ 2mo ago

I don't think this goal was ever realistic, it was just a marketing pitch to get a foot in the door.

dlahoda 2mo ago

It still feels python enough.

melodyogonna 2mo ago

It is still a goal. Mojo's semantics are still driven by Pythons, and can only deviate where safety or performance is an issue. Eventually it'll get the dynamic parts of Python