The Path to Mojo 1.0
16 points by duck_tape
16 points by duck_tape
No mention of Mojo eventually being a superset of Python. Is this still the plan?
Also I am still waiting to program my integrated GPU on my Laptop with Mojo for development to then use the same code on a GPU cloud machine.
This goal was changed ("clarified"?) a bit over a year ago.
Chris Lattner commented in a thread on this topic in the last week, though I'm left a bit confused. On the one hand "it is still a goal for Mojo to grow into an effective superset of Python" although the footnote suggests the Mojo definition of superset might not be what people are imagining "FWIW, I think that people overindex on what “superset” means. C++ is considered a superset of C even though it isn’t literally source compatible. I believe that Mojo will end up with the same relationship to Python in the fullness of time.".
EDIT: Fixed second link. Thank you @mond for spotting the bad copy+paste.
I don't really think so, and I believe it was never really feasible anyway.
It was on some level a marketing gimmick imo, and the best we'll get is an 80% solution. I think this might end up being the right call: I don't think Mojo will work out if they spend all of their efforts banging their heads against achieving full Python compatibility. You have to cut your losses somewhere.
That said, if we get a 99.5% solution... That might usher in a new age of compiler design. I don't think it's feasible without serious leaps on the compiler backend (and that is afaik a huge chunk of what they're working on).