-
Notifications
You must be signed in to change notification settings - Fork 5.8k
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
[data] Streaming executor fixes #2 #32759
Conversation
Signed-off-by: jianoaix <[email protected]>
if ray.data.context.DatasetContext.get_current().use_streaming_executor: | ||
# In streaming execution of ds.iter_batches(), there is no partial | ||
# execution so _num_computed() in LazyBlocklist is 0. | ||
assert ds._plan.execute()._num_computed() == 0 |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
This is basically disabling this test. How about we just force it to use bulk executor instead?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
This assertion still has value as it checks that streaming execution didn't run LazyBlocklist.
Signed-off-by: Edward Oakes <[email protected]>
Signed-off-by: elliottower <[email protected]>
Why are these changes needed?
.dataset_format()
is hardcoded todefault
in streaming, the numpy is treated as a block type which is not supportedds.take()
in bulk execution (v.s. in streaming this is not handled by LazyBlocklist)#32132
Related issue number
Checks
git commit -s
) in this PR.scripts/format.sh
to lint the changes in this PR.