clean up dependencies requirements

[unkcpz]

The principle of my cleaning process is the packages in the list do not conflict with each other and do not duplicate. For example, if the package is installed by aiida-core dep, it should not be specified again in the list.
As for the jupyter backend packages, there is no need to install then, since

aiida-core > 2 will lead to install numpy~=1.19 therefore numpy~=1.23 is installed.

• numpy~=1.17

The following packages are even in conflict and will override the deps which causes the Jupyter engine to fail, we need to pining version when we find future confliction.

• jupyter-client<7

• nbconvert<6

• aiidalab>=21.11.2 is already pinned in the stack

[unkcpz]

The test failed as expected, it rely on aiidalab/aiidalab-docker-stack#333

[danielhollas]

I wonder was there a specific reason for this constraint? nbconvert <6 Do you know @yakutovicha? Are we risking breakage when nbconvert is updated?

[yakutovicha]

I wonder was there a specific reason for this constraint? nbconvert <6 Do you know @yakutovicha? Are we risking breakage when nbconvert is updated?

I remember nothing about this package. One needs to test...

[yakutovicha]

I wonder was there a specific reason for this constraint? nbconvert <6 Do you know @yakutovicha? Are we risking breakage when nbconvert is updated?

I remember nothing about this package. One needs to test...

btw, there is also #384.

[danielhollas]

Thanks @yakutovicha, I was not aware of that issue. Couple days ago I also opened #392 which is related.

[danielhollas]

I think we should take this step by step.

The following packages are even in conflict and will override the deps which causes the Jupyter engine to fail, we need to pining version when we find future confliction.
jupyter-client<7
nbconvert<6
aiidalab>=21.11.2 is already pinned in the stack

@unkcpz could you open a new PR which would only fix the dependencies for the current Docker stack, listed above? aiida-core[atomic_tools] is a separate discussion.

[unkcpz]

@unkcpz could you open a new PR which would only fix the dependencies for the current Docker stack, listed above? aiida-core[atomic_tools] is a separate discussion.

Sure, updated. I add PyCifRW explicitly since it is required by this package.

[danielhollas]

Thanks @unkcpz. I did a quick test locally and it looks good to me. Approving but we should wait until we release a new docker-stack version so that the tests are passing here.

[unkcpz]

The firefox tests are expected to fail and PR #397 is for it, there are two chrome tests that also fail but I think if there are two passes then the implementation is correct.. I'll merge this and if anything happened I'll fix it afterward.

[yakutovicha]

Sorry, I am late to the party.

• aiidalab>=21.11.2 is already pinned in the stack

I am not 100% sure this is a good thing to do. AiiDAlab could be setup outside of the stack, and AWB does require aiidalab package to function.

[unkcpz]

You are right, and I find the doc build always lack the necessary packages if aiidalab package is not explicitly installed. I add aiidalab>=21.11.2 back in #397 to fix the doc build test.