Conversation
|
This should resolve issues like what we are currently seeing in nightly builds: https://github.com/rapidsai/kvikio/actions/runs/6452705318 |
jakirkham
left a comment
There was a problem hiding this comment.
Thanks Vyas!
Think we want exclusive (instead of inclusive) constraints on numcodecs. Otherwise it pulls in the affected version
|
Whoops 🤦 thanks @jakirkham. Fixed. On the plus side, now we know from this run on the last commit that the issue from the nightly runs reproduces on this branch, so we'll have confirmation that this pin is a fix if CI passes this time around. |
|
No worries. Thanks Vyas! 🙏 Should we add this constraint to the Conda recipe too (given kvikio/conda/recipes/kvikio/meta.yaml Line 72 in 7703b40 Wondering if this might fix the couple CI failures still remaining |
|
What about using |
|
CI is still installing |
|
Yeah was thinking about Think we still need a couple more |
|
Apologies just seeing the discussion above, but yes I made the necessary fix to the conda recipe as well. I would also prefer to use |
|
Fine with me, let’s go with <0.12.0
…On Tue, 10 Oct 2023 at 18.58, Vyas Ramasubramani ***@***.***> wrote:
Apologies just seeing the discussion above, but yes I made the necessary
fix to the conda recipe as well.
I would also prefer to use <0.12.0 for 23.10 given that RAPIDS releases
tomorrow and it would be better to be completely decoupled from the
numcodecs fix.
—
Reply to this email directly, view it on GitHub
<#300 (comment)>, or
unsubscribe
<https://github.com/notifications/unsubscribe-auth/AAH6FQH4ZKVEKFZSAARCFFDX6V5B5AVCNFSM6AAAAAA5ZOPXQGVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMYTONJVHA3TAOJYGY>
.
You are receiving this because your review was requested.Message ID:
***@***.***>
|
|
Thanks all! 🙏 |
numcodecs 0.12 appears to have some breaking changes. I've submitted a fix upstream, but for this release of RAPIDS we'll need to upper bound the version to be safe.
xref: zarr-developers/numcodecs#475