How “experimental” are the new core APIs (graphs/j...
# ask-community
w
How “experimental” are the new core APIs (graphs/jobs/ops)? We’re just integrating Dagster into our infrastructure and are wondering if we should start to use the 1.0 abstractions right off the bat. 🤔
❤️ 3
d
I'm also at the same stage thinking same.
d
similar. I just switched over to the 1.0 abstractions, but all the “experimental” runtime warnings are making me sweat a little….
😂 3
w
Warnings?? Pffft. We just call those “log chatter”.
😂 1
@rex thoughts on this?
r
cc @sandy - these are fantastic changes to use, and like we said in our announcement, they’ll be serving as the foundation for Dagster in 0.13.0, so you’ll get a head-start if you write your pipelines in these new paradigms now 🙂
🎉 1
w
Awesome.
s
@William Reed @Dylan Hunt @Donny Winston - to add, we don't have any changes in particular planned right now to those APIs. If we do make changes, it will be because of user feedback
👍 2
w
Hello @sandy, has there been much testing on multiprocessing with the new abstractions? I seem to be able to run fine on production using the S3IOManager, but not on local with the fs or mem ones.
Errors are like this:
Copy code
_pickle.PicklingError: Can't pickle <function mem_io_manager at 0x7fbf37a4a5e0>: it's not the same object as dagster.core.storage.mem_io_manager.mem_io_manager
s
hey @William Reed, do you have a code snippet we could use to reproduce? definitely a concerning error cc @chris
👀 1
w
@sandy my apologies, I think this was my mistake and not a problem with the new APIs (to your relief)! I was passing a function into a kwarg of a new
Process
when I assumed it was a string. Sorry!
cc @chris bc cool profile photo.
blob cowboy 1
s
that's a relief!
w
That pickling error was a head scratcher for a few minutes though. 🙂