taxa_nest

[bibsian]

Remove "shrink" argument form elastic tab.

Write unit test too.

[AldoCompagnoni]

Shrink removed

[bochocki]

Reopening because 1) elastic_tab() is causing some issues, and 2) it's not clear whether the unit test was written.

In browse(), the call to elastic_tab() was originally placed after the vars section (where the user's input could determine which columns to return). This had the (unintended?) consequence of preventing the user from deciding whether or not to return taxonomy; the column was always there, regardless of what the user specified.

• Moved this call to before the vars section.
• Added "taxonomy" to possible_args() so that "taxonomy" does not fail the err_full_tab() test.
• Added "taxonomy" to the err_full_tab() call; since it's not in the main table, a user calling vars=taxonomy would get an error message.

elastic_tab() also seems to collapse all of the taxonomy columns (e.g. "species", "genus", etc) so they do not appear in the full table, even when the full table is specified. This seems to be the intended function, so for now I've commented all taxonomic names out of the possible_args() function so that things get returned properly.

Finally, something in elastic_tab() seems to be changing the table output from browse() in a way I didn't expect. Calling browse(structure == "size", full_tbl = F) returns a dataframe with dimensions [2824 x 17], but calling browse(structure == "size", full_tbl = T) returns a dataframe with dimensions [54 x 61]. The difference in columns was expected, but the difference in rows seems strange to me. I've isolated it to elastic_tab(), but am not sure whether or not this is the intended behavior.

[AldoCompagnoni]

I shall note that Brad points out a conflict that could turn up in 5 different ways. I provide sample code for these 5 scenarios. I provide sample code where I define two types of vars in browse: one non-taxonomic variable, proj_metadata_key, and taxonomic variables:

1. browse(vars=c("proj_metadata_key")): no taxonomy list provided
2. browse(vars=c("proj_metadata_key","genus"): taxonomy list includes "genus" only
3. browse(vars=c("proj_metadata_key", "taxonomy"), full_tbl=F): "small" taxonomy list provided
4. browse(vars=c("proj_metadata_key", "taxonomy"), full_tbl=T): "large" taxonomy list provided
5. browse(vars=c("proj_metadata_key","genus","taxonomy")): here, "taxonomy" overrides "genus", so that taxonomy includes all the variables of the "small" taxonomy list (default is full_tbl = F).

[AldoCompagnoni]

Here is how attempted to solve this issue.

1. when vars = "proj_metadata_key", browse reports only that column
2. same thing when vars = c("proj_metadata_key","lterid": there is not "list column" with taxas
3. when only one taxonomic variable is specified (e.g. vars = "genus"), browse reports a list column whose name refers to the specified variable (e.g. "genus")
4. when more than one taxonomic variable is specified, then those variables go under a list column named "taxas".

[AldoCompagnoni]

Look up issue #14: if my solution above is considered viable, then we'll close both issues.