Warn when PageRank convergence check is ineffective - #8735

I'm iffy on a RuntimeWarning, not because it's not a valid thing to warn about, but because warnings are very often ignored and extra runtime cruft is not necessarily a good thing. I'd propose to stick with documenting this instead. Something like: Note: For large graphs you really need to think about tolerance levels. And then a 1-line description similar to whats in your warning.

Thanks, that makes sense. I've removed the runtime warning and switched this to a documentation-only note in the public pagerank docstring.

The updated version avoids runtime overhead and keeps the PR focused on documenting the tolerance caveat for large graphs.

I also removed the warning-specific tests since there is no longer a runtime warning behavior to test.

Thanks Ross, the parameter-doc placement is much better. Thanks for tightening the wording and for the review.