-
Notifications
You must be signed in to change notification settings - Fork 19
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
feat: update to latest awkward==2.5.0
#407
Conversation
9ebaa37
to
4c7265d
Compare
lists, | ||
meta=typetracer_array(ak.Array(lists[0])), | ||
meta=typetracer_array(ak.Array(lists[0], attrs=attrs, behavior=behavior)), |
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.
@douglasdavis hmm we might need to add a mechanism for passing a type object and/or enforcing this meta - lists could have any type in lists[1]
, etc.
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.
Yea I see what you're getting at here. from_lists
is meant to be a pretty simple entry point to instantiating a dask-awkward array from non-disk data. I originally wrote it as a simple testing ground when adding the from_map
interface and I felt like it wouldn't be a bad addition to the IO API. My thoughts here are to keep it simple and perhaps add a note to the docstring describing the function as expecting the same type for each list.
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.
Perhaps we can make it a runtime error if the subsequent partitions don't have the anticipated meta?
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.
That seems reasonable to me
There's a change in Awkward 2.4.9 that strictly requires |
attrs
awkward==2.5.0
The conda test will fail until conda-forge's bot grabs the latest awkward so we can ignore that workflow |
@@ -47,6 +47,7 @@ def test_list_with_ints_raise(daa: dak.Array) -> None: | |||
|
|||
|
|||
def test_single_int(daa: dak.Array, caa: ak.Array) -> None: | |||
daa.eager_compute_divisions() |
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.
@douglasdavis can you confirm this is a reasonable fix for the changes to implicit len
computation?
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.
We probably want to avoid eager_compute_divisions
on the fixture since it's long-lived, perhaps add a
daa = copy.copy(daa)
above the divisions compute
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.
Ooh, lovely idea.
Codecov ReportAttention:
❗ Your organization needs to install the Codecov GitHub app to enable full functionality. Additional details and impacted files@@ Coverage Diff @@
## main #407 +/- ##
==========================================
+ Coverage 93.85% 93.91% +0.05%
==========================================
Files 23 23
Lines 3125 3153 +28
==========================================
+ Hits 2933 2961 +28
Misses 192 192 ☔ View full report in Codecov by Sentry. |
@douglasdavis I think this is ready to go. I haven't done a pass of adding new |
We need to add tests, but that's a bigger task. I am thinking towards us merging this to fix things, and do a patch release if we spot anything in our later tests. |
Just doing some last checks actually. |
OK, let's hold this, found some strange behavior. |
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.
I gave this another scan- tests are passing and everything looks good to me. When you think it's ready to merge I say go for it.
Thanks @douglasdavis. I found a bug in Awkward, so we can't fix that here. Yet, it's not a show-stopper. I've bumped our minimum version of awkward to 2.5.0. |
This PR adds support for
awkward==2.5.0
's latest features, includingattrs
and a public typetracer API.