allow easy recoding of data columns about isoreader

[japhir]

You could use dplyr's unite for this.

[sebkopf]

good idea, I think what unite doesn't do well though is deal with NA - I imagine a scenario where you have e.g. info x from one type of IRMS format and info y from another but you want to work with them as a single id column: data_frame(x=c(NA, "a"), y = c("b", NA)) %>% unite(id, c(x, y)) creates NA_b and a_NA.

Maybe just a case_when would be easiest, that way the order the columns are provided also determines order of precedence. Maybe something like this with x, y, z dynamically implemented using the unite syntax?

data_frame(
 x=c(NA, "xb", "xc"), 
 y = c("ya", NA, NA), 
 z = c("za", "zb", "zc")) %>% 
mutate(
 id = case_when(
  !is.na(x) ~ x, 
  !is.na(y) ~ y, 
  !is.na(z) ~ z, 
  TRUE ~ NA_character_))

[japhir]

Hmm in that case I don't understand what you want to achieve. This would always pick the first that has a non-na value and ignore further columns, right?

# A tibble: 3 x 4
  x     y     z     id   
  <chr> <chr> <chr> <chr>
1 NA    ya    za    ya   
2 xb    NA    zb    xb   
3 xc    NA    zc    xc   

[japhir]

This discussion on tidyr's unite may be useful (on how to handle NA's). Maybe just replace the NA's with "", then unite the columns?

[sebkopf]

actually implemented this with regular dplyr rename but executed for each isofile separately so recoding to the same column name is easy

[sebkopf]

fully implemented in dev version with iso_rename_file_info() and iso_select_file_info() and supported in all iso_get... functions.