`eurosat_dataset()` might be improved in code style and data design

[cregouby]

Hello,

Current situation

The function could have improved code style

[eurosat_dataset()](#122 ) has a code styling that inherits from other language and prevent readability and portability :

• use of sprintf() instead of glue:: styled message concatenation
• use of base R warning() and if (){stop()} instead of package functions runtime_error("") and value_error("")
• reuse existing message for consistency with other _dataset() functions and preventing specific translation of messages. Example is "Dataset not found. You can use download = TRUE to download it." , ...

The function could have improved data movement design

[eurosat_dataset()](#122 ) uses it's own data folder for archive caching instead of reusing the common torchvision cache folder.

Expected situation

• maximize reuse of existing messages,
• utils.R message functions, and consistent code-style with other _dataset() functions
• reuse the download_and_cache() function for consistent large file storage on end-user machine

[Prateek0xeo]

@cregouby can i update the Utils.r to adapt download_and_cache() for better consistency across dataset loaders?

[cregouby]

Hello @Prateek0xeo,
No, I guess the things to be done are the one in the explicit list in Expected Situation.

[Prateek0xeo]

Hello @cregouby, I sent you a draft of my GSoC application via email but haven't received a response yet. Could you please check and let me know your feedback? Thanks!

[Prateek0xeo]

hello @cregouby, i have been trying to fix the data movement design problem and it seems like the download_and_cache function can not contain a query string (i.e., ?download=true), which is not compatible with file paths on Windows
https://huggingface.co/datasets/torchgeo/eurosat/resolve/main/EuroSAT.zip?download=true
this is the URL i am using, can you help me with this?

also can you also please tell me if there is anything left with code style improvements.