Order of arguments in read_resource

[fjuniorr]

I've noted that the order of the arguments have changed in read_resource from (package, resource_name) to (resource_name, package) in ba943b1.

One of the advantages of the previous ordering is that we could write code with both base (|>) and magrittr (%>%) pipes in the following style:

package |> read_resource("resource-name")

With the new order, with magrittr pipes we can still use . as placeholder

# still not as pleasant
package %>% read_resource("resource-name", .) 

but with the base pipe operator I need

# not fun
package |> (function(x) read_resource("resource-name", x))() 

I couldn't find the reasoning behind the change, but wanted to chime in in the discussion for the design of frictionless-r.

[peterdesmet]

Hi @fjuniorr, thanks for the suggestion!

The switch was initially suggested by my colleague @damianooldoni when thinking about write functions. He found read resource name (1) from package (2) a more natural way of expressing the arguments. I concurred that having e.g. the dataframe first would work well with piping.

However, you make a good case for piping with package. In fact, many of the functions return package:

create_package()                        -> package
create_schema(df)                       -> schema (more a helper function)
get_resource(resource_name, package)    -> resource (more a helper function)
get_schema(resource_name, package)      -> schema (more a helper function)
read_package(file)                      -> package
read_resource(resource_name, package)   -> df
remove_resource(resource_name, package) -> package
write_package(package, directory)       -> package

Some, in addition to:

read_package() %>% read_resource("resource-name")

The following scenario will also supported if we make package the first argument:

create_package() %>%
  add_resource(df1) %>% # function to be developed
  add_resource(df2) %>%
  write_package("directory")

So, I'll revert the switch and make package the first argument where we can. Supporting piping also aligns well with:

Frictionless is an R package to read and write Frictionless Data Packages in a Tidyverse way.