Closed henryiii closed 2 years ago
Aha, this is why this is now breaking:
https://github.com/jupyter/nbformat/pull/238
It was assumed to be pulled in via nbformat, even though there's direct usage of it within this package. Any package directly used should be directly depended on. ;)
Looks like https://github.com/jupyter/nbformat/releases/tag/5.2.0 came out two hours ago.
ipython_genutils is old as dirt. I wonder how quickly it can just be removed from nbconvert?
https://pypi.org/project/ipython_genutils/#description says:
Pretend this doesn’t exist. Nobody should use it.
@minrk , @amueller, @mpacer, and @t-makar are the committers of the imports that regard ipython_genutils
as observed in https://github.com/jupyter/nbconvert/search?q=ipython_genutils
...@minrk, #t-makaro, and @wolever are the committers using the imported modules of ipython_genutils
, path.ensure_dir_exists
, path.link_or_copy
, and text.indent
how quickly it can just be removed from nbconvert?
How quickly can it? PRs welcome!
are the committers of the imports
Calling out folks that have been contributing and maintaining these projects over the last ten years doesn't really help much.
Given this is breaking a lot of readthedocs builds (almost all that use nbsphinx?), quickly adding the dependency, making a quick release, and then working on removing it would probably be a lot better. I guessed removing it was probably the "better" solution since nbformat removed it, but nbconvert does use it right now. But it doesn't declare so.
I think some jupyter_releaser workflows are also affected (e.g. https://github.com/mamba-org/quetz/pull/495)
how quickly it can just be removed from nbconvert?
How quickly can it? PRs welcome!
are the committers of the imports
Calling out folks that have been contributing and maintaining these projects over the last ten years doesn't really help much.
Sorry, i could have shared my reasoning: if these folks recall why these tools are used, and are able to replace or remove them, that's surely one of the quickest ways to resolve this. I certainly do not intend to "blame" anyone, and am truly grateful for their contribution over the years that has made this package one that we are able to use!
My apologies for any misunderstanding.
Many (almost all) of the jupyter projects were once in a single repo/package, ipython/ipython, and then split into many repos and packages. Most of them depend on each other, in one way or another, and all of those need attention, and many of us are kinda tired.
As to ipython_genutils
: if it ain't broke, it doesn't get fixed, and ipython_genutils
just keeps on trucking. It's really only with the final end-of-life of nose
that it's started to no longer be tenable for downstreams to package, and we have been getting trickles of contributions to replace its functionality in different places.
@bollwyvl https://github.com/jupyter/nbconvert/pull/1727 removes the dependencies by bringing the needed functions into this repo instead of adding a dependency to an unmaintained package
Recently I had readthedocs break with
No module named ipython_genutils
coming from nbconvert. Looking at the code in nbconvert, I clearly seeipython_genutils
used in a few places, but no dependence onipython_genutils
declared in setup.py.RtD log: https://readthedocs.org/projects/hist/builds/16329065/
Inline log error:
```pytb Running Sphinx v4.4.0 loading translations [en]... done Traceback (most recent call last): File "/home/docs/checkouts/readthedocs.org/user_builds/hist/envs/387/lib/python3.8/site-packages/sphinx/registry.py", line 433, in load_extension mod = import_module(extname) File "/home/docs/checkouts/readthedocs.org/user_builds/hist/envs/387/lib/python3.8/importlib/__init__.py", line 127, in import_module return _bootstrap._gcd_import(name[level:], package, level) File "Manually adding this to my requirements fixes the build. https://github.com/scikit-hep/hist/pull/387/commits/611c452c88866131acb1080b3c0f821aa29f61d7
Nbconvert version: 6.4.2