More tips about how to reduce resources usage #4419
Conversation
| ----------------------------------------------- | ||
|
|
||
| There are a few libraries that Read the Docs has already installed (scipy, numpy, matplotlib, pandas, etc) | ||
| in the Docker image used to build your docs. You can check the updated list of pre-installed libraries in the `Docker image repository`_. |
There was a problem hiding this comment.
We probably need to document which exact libraries are available. I don't think it's reasonable for people to have to dig through a git repo for this data.
There was a problem hiding this comment.
Do we want to document this here or in the specific YAML config at #4422 ?
Also, this depends on the Docker Image used itself so, I'm not sure that there is a generic way to document this and keep it updated. That's why I mentioned them generally and points to the repository.
There was a problem hiding this comment.
I think documenting this in the yaml config makes sense, or maybe on another central doc, because we'll be having multiple versions, and each with the same content about build images.
| in the Docker image used to build your docs. You can check the updated list of pre-installed libraries in the `Docker image repository`_. | ||
|
|
||
| To use these pre-installed libraries and avoid consuming time re-downloading/compiling them, | ||
| you ca use the :ref:`yaml-config:Python-use-system-site-packages` option to have access to them. |
| --------------------- | ||
|
|
||
| In case you are using `conda` to create your environment, | ||
| you could consider using `pip` instead since `conda` could require too much memory to calculate the dependency tree. |
There was a problem hiding this comment.
I feel like this sentence needs some sort of supporting sentence. Why does conda use more memory than pip? Also, I'd phrase it as "If you don't need conda to create your environment, ..."
There was a problem hiding this comment.
I don't know the exact reason of why this is true, but it behaves like that. Also, there are different issues reported at different repositories about this problem of conda consuming too much memory. An example in the official repo: conda/conda#5003
I agree with you that we need to be more explicit here, but I need to think about how to rewrite it.
| Use pip when possible | ||
| --------------------- | ||
|
|
||
| In case you are using `conda` to create your environment, |
There was a problem hiding this comment.
Use double backtips `` (conda and pip)
| in the Docker image used to build your docs. You can check the updated list of pre-installed libraries in the `Docker image repository`_. | ||
|
|
||
| To use these pre-installed libraries and avoid consuming time re-downloading/compiling them, | ||
| you ca use the :ref:`yaml-config:Python-use-system-site-packages` option to have access to them. |
There was a problem hiding this comment.
The ref should be yaml-config:python.use_system_site_packages #4422
|
Looks good. #4515 should have fixed the listing of things included. |
|
@ericholscher #4422 was needed too! |
Depends on #4418
Related to #4362 (comment)