Warning when using laminr in jupyter notebook and connect message not shown

[ap--]

I get this warning when running laminr==1.2.1 with lamindb==1.17.0, and the

→ connected lamindb: org/instance

is only shown in the terminal jupyter logs

To reproduce, install pixi, use the below pixi.toml and run:

pixi run -e laminr-jupyter jupyter

[workspace]
version = "1.0.0"
name = "lamin-base"
description = "Base environment for working with lamindb Python"
platforms = ["linux-64", "osx-arm64", "osx-64", "win-64"]
channels = ["conda-forge", "bioconda"]
authors = ["Andreas Poehlmann <andreas@poehlmann.io>"]
requires-pixi = ">=0.60"

[system-requirements]
macos = "14.0"

[tasks]
lamin = { cmd = "lamin" }

[dependencies]
python = "3.13.*"
pip = "*"

[pypi-dependencies]
lamindb = { version = ">=1.17.0,<2", extras = ["gcp"] }

# =============================================================================
# Feature: jupyter - adds JupyterLab
# =============================================================================
[feature.jupyter.dependencies]
jupyterlab = ">=4.5.0"
ipython = ">=9.0.0"
ipywidgets = ">=8.0.0"

[feature.jupyter.tasks]
jupyter = { cmd = "jupyter lab" }

# =============================================================================
# Feature: laminr - adds R support with laminr package
# =============================================================================
[feature.laminr.dependencies]
r-base = "4.5.*"
r-laminr = ">=1.2.1,<2"
# r-anndata = { version = "*", channel = "bioconda" }
r-arrow = "*"
r-jsonlite = "*"
r-knitr = "*"
r-nanoparquet = "*"
r-quarto = "*"
r-readr = "*"
r-rstudioapi = "*"
r-rsvg = "*"
r-seurat = "*"
r-testthat = "*"
r-yaml = "*"
r-irkernel = ">=1"

[feature.laminr.tasks._fix_reticulate]
# workaround for https://github.com/rstudio/reticulate/issues/1650
cmd = 'echo "# cmd: $CONDA_PREFIX/bin/conda" > $CONDA_PREFIX/conda-meta/history'

[feature.laminr.tasks]
jupyter = { cmd = "jupyter lab", depends-on = ["_fix_reticulate"] }

[feature.laminr.activation.env]
PIXI_R_LIBS = "$CONDA_PREFIX/lib/R/library"
R_LIBS = "$PIXI_R_LIBS"
R_LIBS_USER = "$PIXI_R_LIBS"
RETICULATE_PYTHON = "$CONDA_PREFIX/bin/python"

# =============================================================================
# Environments
# =============================================================================
[environments]
default = { solve-group = "default" }
jupyter = { features = ["jupyter"], solve-group = "default" }
laminr = { features = ["laminr"], solve-group = "default" }
laminr-jupyter = { features = ["jupyter", "laminr"], solve-group = "default" }

[lazappi]

I think this is a general problem with using {reticulate} in Jupyter rstudio/reticulate#1049.

This works in RStudio:

But not in Jupyter:

It seems that this happens because Jupyter only shows R output (not Python anything else) when using the R kernal IRkernel/IRkernel#416, so I think it's actually a Jupyter issue.

I'm not sure we can do much about it here. In theory we might be able to capture and repeat all the Python output but that would require significant changes and when I have tried similar things before it has turned out to be quite difficult.

[falexwolf]

Given most people who use R use RStudio and not Jupyter this might be not that big of a problem.

If many people use Jupyter Lab, we could at least show a few hand-crafted logging messages; but it'd be difficult to then not duplicate with other R environments like RStudio or simple Rscript. So, this is indeed a tricky problem to do well and would ideally be resolved upstream in Jupyter and reticulate.

[ap--]

Hmm, that's unfortunate. But I agree, that it's an upstream issue, and likely a uncommon environment for R users. Would it still be possible to detect if running in an jupyter notebook, and on library(laminr) or laminr::import_module print a warning that lamindb log output won't be displayed due to missing support in IRKernel?

[lazappi]

I've opened a draft PR that does this. @ap-- could you please check this does what you expect? You should be able to install the branch with remotes::install_github("laminlabs/laminr@issue-243/warn-on-jupyter").