moldach / vaporwave Goto Github PK

My extreme frustration with trying, and failing, to build open source projects with fonts and reticulate on Windows OS inspired this issue 🪟🙏🏼

I would like to share some insights and observations from my recent endeavors to optimize Docker images for our RStudio projects, specifically those based on r-ver:4.3.2 and its derivatives.

In the quest to streamline my Docker images, I employed two prominent tools: slimtoolkit/slim and goldmann/docker-squash. Here are the key findings from this exercise:

• The slim tool, while significantly reducing the image size to approximately 150MB, aggressively removes essential components, including s6-env and other critical elements necessary for the rstudio image to function correctly. Consequently, the resulting image, despite its reduced size, fails to operate as expected.

• The docker-squash tool takes a more conservative approach, yielding a modest size reduction of roughly 1-2%. This reduction is consistent across both r-ver, rstudio, and vapoRwave images. Thus, it appears that docker-squash may not effectively optimize R packages, suggesting that the onus of optimization lies with the developer.

• Upon further analysis, I identified that my vapoRwave package/container contributes an additional 1GB to the image size, accounting for one-third of the total image footprint. It is apparent that initial optimization efforts should target this package.

Moving forward, I propose to focus on refining the vapoRwave package to minimize its impact on the image size. Once complete, I will revisit the overall Dockerfile optimization to enhance efficiency while maintaining the necessary functionality of the RStudio environment.

I will use the following tools and procedures to reduce the size of vapoRwave v0.2.0:

• tools:package_dependencies()
• remotes::system_requirements()
• miniCRAN
• Avoid using Depends: in DESCRIPTION:
  • Use Imports: for required packages
  • Use Suggests: for testing, documentation, or optional features
• Use NAMESPACE instead for importing packages/functions (import() and importFrom() respectively)

Hi, I'm trying to get this package onto conda forge. Can a release on Github be made to facilitate that?

Including the fonts in inst/ would make these fonts accessible to R code via system.file() -- allowing them to be be registered and loaded via extrafont within the vapoRwave package.

The DESCRIPTION file would need to reflect the font authors, copyright holders, etc. Most of the fonts seem to not have any license restrictions prohibiting this kind of distribution.

There are 13 scales however currently there are only 3 themes.

Generate a few more themes.
Maybe also add a helper-function to make creating new themes + scales for PRs easier