Explain column-independent operations

[bkamins]

This PR adds additional explanation of context dependent expressions in operation specification syntax to make sure users know exactly how they work.

CC @yjunechoe @pdeffebach

[bkamins]

I additionally expanded a section on indexing and iterating a grouped data frame.

[nalimilan]

Thanks! I've made many small comments.

I wonder whether "context-dependent" is explicit enough. The operation specification syntax is already context-dependent as it doesn't work outside of combine/select/transform. And any function call can be thought as context-dependent as the result can vary a lot depending on what columns are present in the data, on whether a DataFrame or a GroupedDataFrame is passed... Not sure what a better term could be: maybe something which stresses that these look like function calls but are actually not? "pseudo-functions"? "dummy functions"? mock-functions"? Maybe there's already a term for this kind of pattern in other languages?

[bkamins]

I wonder whether "context-dependent" is explicit enough.

This term is used in dplyr https://dplyr.tidyverse.org/reference/context.html.

I tried to think about a better name, but I really did not find a great proposal. I will ask on Slack for comments.

[bkamins]

The alternative names proposed on Slack are:

• column-independent operations
• column-independent expressions
• context-aware operations
• non-column sources (or non-column operations)

[bkamins]

The PR is updated to use "column-independent operations" as it seems the least controversial option.

[bkamins]

@nalimilan - OK to merge this PR? Thank you!

[nalimilan]

Still relevant.

[bkamins]

Still relevant.

@nalimilan - what do you mean here? Could you please clarify?

(GitHub does not show me what you refer to). You wanted me to put together all examples with printing number of rows for all groups in one place and I did this.

[bkamins]

Thank you!