You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
The fiagnos_docs.ipynb has a handful of cells that really take a long time to render properly. This is due to the fact that there are calculations being run to generate ensemble dataset outputs using xclim.ensembles that are also reliant on the PAVICS THREDDS data node. This is way too heavy to reliably run on ReadTheDocs or locally, which explains why there are timeout errors (depending on network connectivity).
One calculation in particular (the setup code below "Plotting Hatched Areas") can take several minutes to run (more than 10 mins on my machine). If on-the-fly calculations are needed for this, the example needs to be limited to a smaller ensemble over a narrower period/area.
Steps To Reproduce
$ make docs
Additional context
If we want to ensure that figanos passes xclim integration tests to ensure that the API remains compatible, we should try an approach that stresses that using pytest:
Take note of the expected outputs of xclim when creating ensembles.
Write a handful of tests to ensure that the required metadata fields are present using a calculation that relies on a very limited time period and domain.
Assert that this info is parsable by figanos
For the documentation:
We should be making use of the xclim-testdata repository and place the required ensemble files there, to be fetched using the xclim.testing.utils.nimbus utility.
We should split the documentation into a few more pages so that we can potentially make use of distributed documentation builds (faster).
Contribution
I would be willing/able to open a Pull Request to address this bug.
The text was updated successfully, but these errors were encountered:
Setup Information
Description
The
fiagnos_docs.ipynb
has a handful of cells that really take a long time to render properly. This is due to the fact that there are calculations being run to generate ensemble dataset outputs usingxclim.ensembles
that are also reliant on the PAVICS THREDDS data node. This is way too heavy to reliably run on ReadTheDocs or locally, which explains why there are timeout errors (depending on network connectivity).One calculation in particular (the setup code below "Plotting Hatched Areas") can take several minutes to run (more than 10 mins on my machine). If on-the-fly calculations are needed for this, the example needs to be limited to a smaller ensemble over a narrower period/area.
Steps To Reproduce
$ make docs
Additional context
If we want to ensure that
figanos
passesxclim
integration tests to ensure that the API remains compatible, we should try an approach that stresses that usingpytest
:xclim
when creating ensembles.figanos
For the documentation:
xclim-testdata
repository and place the required ensemble files there, to be fetched using thexclim.testing.utils.nimbus
utility.Contribution
The text was updated successfully, but these errors were encountered: