simone viozzi

Same thing, on the nuclio dashboard, there are some logs: ```text [13:16:32.906] (I) Deploying function [13:16:32.906] (I) Building [versionInfo: "Label: 1.5.16, Git commit: ae43a6a560c2bec42d7ccfdf6e8e11a1e3cc3774, OS: linux, Arch: amd64, Go version:...

torch 1.12.1+cu116 hiddenlayer 0.3 same error with code: ```python # VGG16 with BatchNorm model = torchvision.models.vgg16() # Build HiddenLayer graph # Jupyter Notebook renders it automatically hl.build_graph(model, torch.zeros([1, 3, 224,...

Any idea of a roadmap to fix it? With this broken https://github.com/iterative/PyDrive2/pull/222 is pretty much useless, it will work unexpectedly 90% of the time. I could try, but I still...

`copy` uses `expand_path`, which uses `find`. https://github.com/fsspec/filesystem_spec/blob/2633445fc54797c79a7ac96d213bd24dfbdfcdc2/fsspec/spec.py#L877 https://github.com/fsspec/filesystem_spec/blob/2633445fc54797c79a7ac96d213bd24dfbdfcdc2/fsspec/spec.py#L908 `cp_file` is fine, though, but works for just files.

Hey @shcheklein, did you know that `find` is already implemented by `ffspec` and it uses `walk`: https://github.com/fsspec/filesystem_spec/blob/2633445fc54797c79a7ac96d213bd24dfbdfcdc2/fsspec/spec.py#L423-L453 And also `walk` is already implemented by `ffspec` using just `ls`: https://github.com/fsspec/filesystem_spec/blob/2633445fc54797c79a7ac96d213bd24dfbdfcdc2/fsspec/spec.py#L364-L421

@shcheklein I noticed that there was no `mkdir` method and I needed one, so I implemented it. I also noticed that `fs.expand_path` only works if the path is one level...

Right now is possible to rename a file using existing methods: ```python file.FetchMetadata(fields="title") file["title"] = "new_file1" file.Upload() ``` and looking at the code for `upload`: https://github.com/iterative/PyDrive2/blob/64a50a1580b2d276f8c9563f5ba620f7e7706119/pydrive2/files.py#L499-L512 it uses `_FilesPatch` internally....

> is it possible to do this w/o a call to Fetch? (if I know the file id). I wonder also if it's possible to do this if I know...

oh, i get it now, thank you! i think we can definitively optimize them, maybe for the rename we could use `self.auth.service.files().patch(**param).execute()` directly, so we don't need the fetch before...

> The most annoying thing on Google Drive is that one path can resolve into multiple IDs. check out: https://rclone.org/commands/rclone_dedupe/ i think for the move / rename method we can...