Solutions for a crowded namespace: selective removal of items? Any better ideas?

[vertesy]

Hello,

I use tidyverse, and bc its a dependency for many packages, frankly, it's not really a choice.

The problem

The global namespace often gets crowded with lot's of unneccesary variables.

This is a problem with autocomplete, bc rarely used items, often takes precedence over key base functions, for example

• typing ta and hit autocomplete should result in table(), but once I load tidyr the default complete becomes table1 which is some random example datset
• Another example r is pr autocompletes to presidential instead of print().
• I believe most people use these example datasets much much less frequently then core functions.
• Such namespace crowding makes autocomplete often useless, thus shaving off minutes every working hour.

---

Examples

Looking for a solution

1. Not loading tidyverse packages is not a reasonable option
2. I cannot remove by rm(table1), it stays
3. I can manually overwrite a long list of these with table1 <- NULL, which makes it disappear from the autocomplete list, but they stay in global namespace
4. How would you deal with this? Do u have a better solution?

What would be a better long term solution

1. I think many package creators are lenient and spam the workspace (I am no exception). At least the separation of data objects should be loadable independently, such as done for Seurat and SeuratData. For a lack of a better idea, I would like to propose that you separate tidyr and tidyrData, or tidyverseData into separate packages. I use tidyverse every day, I don't think I ever used these data objects.

Thank you for your time!

---

ps. you can see another bug on the screenshot: displayed package assignment is often wrong (in RstudioServer Desert Rose): prcomp is from stats and not from MarkdownReports.

[DavisVaughan]

I don't have a great solution but I don't think there is really much we can do about this. I'm not sure it is reasonable to think that typing in pr should always result in print() - I think it is reasonable for package authors and IDE developers to hope that you will type in a few more characters to narrow the search further.

I also don't think we can move the data to another package at this point. Teachers often rely on these example data sets being loaded alongside the package.

[vertesy]

Ok, well that's disappointing, but life isn’t a cakewalk...

My first example acutely pinpoints that typing any number of characters for table() can result in weird auto completion, so hanging onto pr... is beside the point.
I think the auto-completion feature changed in the meanwhile for better ordering.

Many other packages separate data from code, eg Seurat and seurat-data (ofc you can make the case that it's big data). It is just an example.

I guess "teachers" (people) would need to do: require(tidyr); require(tidyr_data). Selective removal may be an alternative.

May I ask, how many data objects are loaded automatically in the environment by tidyr and tidyverse?