Main question:
What's the best way to pass variables to dplyr and data.table - when variables are not hard-coded but are acquired from user at run-time, as when entered from Shiny?

Example 1:

df <- mtcars
dt <- mtcars %>% as.data.table

input <- list(); input$newcol="rate"; input$numerator="cyl"; input$denominator="wt"

df <- df %>% mutate(get(input$newcol)=get(input$numerator)/get(input$denominator))
dt[ , ':='(get(input$newcol)=get(input$numerator)/get(input$denominator))]

Example 2: (The same as example 1 but referencing columns by number)

input$numerator=2; input$denominator="6"

Example 3 (The same as examples 1 and 2, but using array of variables ):

input$newcols=paste("rate_", 1:3); 
input$numerators=1:3 
# or input$numerators=c("cyl", "disp", "hp"); 
input$denominators=5:7

And tangent questions:
Q1: What is wrong in the example 1 above? (the error message is shown below)
Error: unexpected '=' in "df %>% mutate(get(input$newcol)="

Q2: Can/Should one use as.name() instead of get() ? When one should use the former, and when the latter?

PS: I'm suprised this is not normally discussed anywhere in early stages of R learning books (including from Harley and Matt Doyle). If I would teach R, this would be the first skill I would teach - how to write code that it is reusable with any variable (not with just hard-coded variable), like we teach C++ or Java. Comment on this PS are also welcome.