James Lamb

That could definitely be related, especially if something on the `dask`-specific code paths leads to conflicting shared-library-loading (like some of the issues we've talked about in #5106).

I was able to reproduce this in Docker tonight, using the same image and environment variables as the `Linux bdist` job. ```shell # Linux bdist docker run \ --rm \...

I tried running the tests with `LD_DEBUG=libs` set ([see here for docs on that variable](https://man7.org/linux/man-pages/man8/ld.so.8.html)), to see what was being searched. Like this. ```shell LD_DEBUG=libs \ pytest \ -v \...

@StrikerRUS sorry it took me so long to respond to this! I agree with the proposal to bump the minimum `glibc` version and to start publishing wheels with using one...

> start publishing wheels with using one of the standard `manylinux` images @StrikerRUS I just returned from traveling and have some time to work on this. But I saw that...

> Because the fix is urgent for a usage in our product team. This is an open source project, not a Microsoft commercial product. I feel strongly that choices we...

Ah ok thanks @shiyu1994 , now I understand. I guess the phrase "When the input dataset is from LibSVM file" was used in #5156 because loading from a file bypasses...

I can review by the end of this week. Sorry for the delay @jgiannuzzi , and thanks for the awesome work!

I'll be able to review today, sorry I didn't get to it yet!

Thanks for that! I'll take a look shortly. In the future, when contributing on this project please use merge commits instead of rebasing + force-pushing. Rebasing and force-pushing makes it...