Unable to rebuild working datascience-notebook? #668
Comments
|
Yup, I'm having the same issue with R |
|
We try to pin the minor versions of packages listed in the dockerfiles here. If new patch releases or new builds of those packages become available between builds, the results will not be the same. If the packages we list do not pin their dependencies similarly (and most do not), it's possible for those dependency packages to change across builds as well. PNG support for R has been a problem here in the past (e.g., #210). It's possible there is a compatibility problem again among R, IRkernel, libpng, etc. |
parente
added
type:Bug
|
I believe #679 has the links upstream to the R package problems from conda-forge that are causing this problem. I'm going to close out this issue as a duplicate since the other has more direct information about the problem. |
parente
added
status:Duplicate
What docker image you are using?
Example:
jupyter/datascience-notebookWhat complete docker command do you run to launch the container (omitting sensitive values)?
Example:
docker run --rm -it -p 8888:8888 jupyter/datascience-notebook jupyter labWhat steps do you take once the container is running to reproduce the issue?
Example:
capabilities()What do you expect to happen?
Expect to see that e.g. PNG support is enabled
What actually happens?
PNG support (and others) not enabled.
I'm trying to trouble shoot some issues that have arisen since the docker-stacks started using Bionic 18.04. In the process of that, I'm trying to re-build the datascience-notebook using the jupyter/docker-stacks repo. I go through building base-, minimal-, scipy- and then datascience- using "make build/base-notebook" etc on a Mac.
When I run R notebooks through
docker run --rm -it -p 8888:8888 2339546dfd7b jupyter lab, the container logs complain:and when I check the
capabilities()in the installed version of R, I see that PNG is reporting FALSE (i.e. no PNG support). If I pull the repos from 16 days ago from dockerhub, R works.But, if I start with a clean Docker system (did a reset, etc) and do a "build-all-test" and test the resulting datascience-notebook, I get the results above.
I can't figure out what is breaking or the difference between the two because the version at https://hub.docker.com/r/jupyter/datascience-notebook/tags/ appears to have been built from the June 23rd commit that would be the most recent commit in my clone.
Any ideas why building on docker locally vs. using the dockerhub image would produce different results?
The conda packages installed are shown below.
The text was updated successfully, but these errors were encountered: