ROADMAP 2022/2024

[RMHogervorst]

Hi fans, and contributors to {charlatan},

I'm Roel (@RMHogervorst on github) and I'm taking over the maintenance of this package.
I'm using this issue to sketch out a generic roadmap for the development of this package.

In general (and looking at the open issues) I'm seeing a few themes:

1. Adding more features (providers and locales) to match perls faker and python 's faker functionality.
2. improve docs and examples to show what is already possible with the current package.
3. making it easier to extend the charlatan package (making new locales and new providers)
4. make it easier to maintain.

Charlatan is set up in a very modular way, and that is awesome!
My primary goal right now (september 2022) is a major refactor in the spirit of #116 , making classes for every locale.
I'm convinced that will make it easier to extend the package (3) and easier to maintain (4)

Roadmap

This roadmap shows in what order I want to work on improving/maintaining the package. This is a rough guide, and I welcome any help you can give me.

• refactor existing code to use classes in stead of dynamic loading (3, and 4)
• [ ] showcase all the current functionality (2)
• [ ] performance improvements
• [ ] adding new features (1)
  If you have different ideas or want to contribute on features you are absolutely welcome to open an pull request or issue!

refactor existing code to use classes in stead of dynamic loading

Why? Locale specific providers have different needs, right now we have a massive class with many functions that are not necessarily useful for all locales. Examples from addresses:

• there are military base specific addresses in us
• spanish and french (all latin languages?) streetnames form {{street_prefix}}(de) {{first_name}} {{lastname}} Rue de Charles Gaule, while English (and german and dutch, and belgium) streetnames often form like {{first_name}} {{last_name}} {{street_suffix}}: Charles Darwin Street.
• Different administrative boundaries. In the USA you would use States as a designation Cambridge massachusetts, but in Dutch we almost never use the province.

Using inheritance we can re-use the basic functionality per provider but make it more custom for locales, do you need a specific function to create parts of your address? that is now possible inside the localized class.
What will this enable? This should make it easier for contributors to add a localized provider for their language. It should also make it easier to port functionality from python faker.

• work in issue Re-arrange locale specifics for each provider #116 and branch feat/rearrange_locale

showcase all the current functionality

why? The getting started and README only show some functionality, but there is way more! And I personally found it hard to find docs or examples. We can do better!

• create a vignette to to show all the current functionality
• create blogposts/ vignettes to show how you can create realistically looking data with correlated structures, by using several packages together.

performance improvements

Creating many values is rather slow, possibly because of whisker, possibly because we create single values, possibly because of the way we call functions everywhere. Fact is that creation is rather slow and I don't yet know why. Instead of guessing we need to benchmark and profile the code.

• benchmark current package creation times for 1, 10, 100, 1000, 10000 of a function.
• profile current package data generation
• use lessons learned from profiling to improve performance

Adding new features

There are plans in the issues for

• adding a misspelling ability Generate typos and error #86
• social security number provider Feature request: Add random U.S. Social Security Numbers #121
• randomly assigned values from user supplied categories new fxn/provider idea: vector containing randomly assigned values from user supplied categories #105
• barcodes, isbn codes Various codes #56
• product types, prices, order ids fake business transactional data #52
• ILO, ISCO, ICSE status in employment, occupation classification

[RMHogervorst]

Updated Roadmap with closed issues and next steps