-
Notifications
You must be signed in to change notification settings - Fork 31
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
Set sharktank package deps during package build reliably. #440
Conversation
We could mirror these between requirements.txt and pyproject.toml or keep them dynamic. For packaging, we don't want to set explicit versions unless absolutely required. Developers may want to pin to a specific version to make testing consistent.
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 could mirror these between requirements.txt and pyproject.toml or keep them dynamic. For packaging, we don't want to set explicit versions unless absolutely required. Developers may want to pin to a specific version to make testing consistent.
@stellaraccident opinions?
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.
Industry guidance:
-
https://packaging.python.org/en/latest/guides/writing-pyproject-toml/#dependencies-and-requirements
-
https://setuptools.pypa.io/en/latest/userguide/pyproject_config.html#dynamic-metadata
Library packagers are discouraged from using overly strict (or “locked”) dependency versions in their dependencies and optional-dependencies.
@@ -1,7 +1,12 @@ | |||
iree-turbine | |||
|
|||
# Runtime deps. | |||
gguf==0.6.0 | |||
numpy==1.26.3 |
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 ran into errors trying to unpin these versions (replacing ==
with >=
). Sample logs: https://github.com/nod-ai/SHARK-Platform/actions/runs/11714432277/job/32629101709#step:6:64
Maybe one of the deps doesn't support numpy 2.0? I can play dependency-updates-whack-a-mole to try unpinning some of the packages I guess...?
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 need to, yes :/ Can help with that on Monday. Should probably fill a follow up issue to keep track.
@@ -1,7 +1,12 @@ | |||
iree-turbine |
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.
As part of this I added a dep on
iree-turbine
to sharktank'srequirements.txt
and then switched workflows and user guides to installsharktank
first and then install a newer/local version from git if desired.
@stellaraccident is this set of changes okay?
@@ -1,7 +1,12 @@ | |||
iree-turbine | |||
|
|||
# Runtime deps. | |||
gguf==0.6.0 | |||
numpy==1.26.3 |
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 need to, yes :/ Can help with that on Monday. Should probably fill a follow up issue to keep track.
Co-authored-by: Marius Brehler <marius.brehler@gmail.com>
887fb77
to
7ee48ae
Compare
Progress on #294.
We could mirror requirements between
requirements.txt
andpyproject.toml
or keep them dynamic. For packaging, we probably don't want to set explicit versions unless absolutely required. Developers may want to pin to a specific version to make testing consistent though.As part of this I added a dep on
iree-turbine
to sharktank'srequirements.txt
and then switched workflows and user guides to installsharktank
first and then install a newer/local version from git if desired.