HN Explorer

Anyone Moving from Julia to Mojo

1 points by mrsofty ↗ HN
we've tried using julia but FINALLY the compile time latency, HUGE memory consumption, language instability and pkg version hell got to us. We're lucky that we modularized everything so we can easily move over in stages. I'd be interested in anyone else who are making the move and what field they are in. For the record julia is a fine language but it is NOT for us.

14 comments

[ 3.5 ms ] story [ 48.3 ms ] thread
ftxbro
Mojo isn't ready unless you are doing something like writing for-loops in Python and you want them to be fast. It has probably even more language instability than Julia and it doesn't have a package manager so I don't know if that's better or worse than pkg version hell. There hasn't been enough time for multiple package versions though, so there's that.
mrsofty
we have found a few other things that it does.

https://docs.modular.com/mojo/notebooks/Matmul.html https://docs.modular.com/mojo/notebooks/BoolMLIR.html

ftxbro
well that's an implementation of matmul as three nested for-loops and making it fast. so it's exactly "writing for-loops in Python and you want them to be fast." it's not a other thing.
mrsofty
fair point how about this https://docs.modular.com/mojo/programming-manual.html

AND you missed the second link in the post https://docs.modular.com/mojo/notebooks/BoolMLIR.html

ftxbro
you just posted the entire mojo programming manual
mrsofty
I did, to highlight the fact that it does more than you observed.
ftxbro
I mean I didn't say that mojo only has for-loops. I said that speeding up for-loops is the only reason to use it.
npalli
Dude, six months back you were fed up with Julia and wanted to move to Carbon . How did that work out? Now you want to move to Mojo?? Who’s funding any of these attempts or are you just screwing around with some small personal scripts.

https://news.ycombinator.com/item?id=35009074

mrsofty
Yes, that's when it started. Then we sent some people to Juliacon '23 to get a feel for the future. Julia is a fine language, just not for us given the reasons in the OP.

We have been tracking mojo since 2020 but wanted to wait for a release date before making the change. Dr Lattner has a reputation of producing commercial quality work and we trust that. Also, because our code base is modular by design ( Dr Jim Gray was our mentor), and bound together by ipms ( any transport will do), we can make the conversion very quickly. As to funding it's mostly us. We have done reasonably well in financial trading augmentation approaches and so the funding is internal. We love writing small personal scripts in any language and much of our work product has been small and reactive, I suppose you could call them scipts. I would point out that NONE of this is relevant to my OP question.

mrsofty
Hi there I am mrsofty, the OP. I just want to point out that the question is

" I'd be interested in anyone else who are making the move and what field they are in?"

I phrased it like that to attempt to keep it to simple. We have made the decision based on 6 months of internal discussion, compiler and hardware expert input and watching the roll out of Mojo chatter. I just wanted to clear that up.

DNF2
I'm curious about the "language instability". What kind of stability is that? Are talking about package instability, or the core language?