Closed scarlehoff closed 6 months ago
I will merge this by the time https://github.com/NNPDF/nnpdf/pull/1861 is implemented.
If someone (@comane I guess is the only person with experience here) can review before that so much the better.
Since this is only dealing with a few test and dependencies and the latest version is already being used via conda, I guess it would be fine.
@comane when you have time, could you please review this?
@scarlehoff could you add a comment explaining why you restricted the ruamel_yaml
version?
One of the reasons this needed review was because I did the things that made the code work.
Including ruamel. This version worked, others didn't...
@comane are you fine with me merging this?
afaik you are the only one using the newer versions of validphys
Hi @scarlehoff, I tested this branch with the most up to date version of validphys (nnpdf), by pip installing this branch. I runned some example scripts (plot pdfs and data theory comparison) and they worked well even with dask.
I guess that the idea is to change the nnpdf dependency as well since right now it is checking a fixed commit (which also needs curio).
I added a suggestion for bokeh. This way one gets the dask dashboard.
I see. I'll add bokeh then as an extra. Since dependencies of reportengine affect nnpdf as well I'd rather not add many.
I guess that the idea is to change the nnpdf dependency as well since right now it is checking a fixed commit (which also needs curio).
Yes. At least for python 3.12. But the first step is to create a conda package for this (since that's the blocker right now)
This PR does:
updates the
ruamel
anddask
dependenciesModify tests in
test_executor.py
andtest_builder.py
so as to be able to accept dask futures.