F() overriding FALSE

[edrubin]

First off: Thank you so much for a fantastic package.

I keep running into a problem generated by the F() function: it has overwritten base R's logical vector (for FALSE). Is there any way we give F() a different name? I know there's an issue of backward compatibility, but a lot of my previous code also breaks when collapse::F() masks base::F.

Simple example of the problem

library(data.table, collapse)
# Create a data table
> dt = data.table(var = c(TRUE, FALSE), val = 1:2)
# View data table
> dt
     var val
1:  TRUE   1
2: FALSE   2
# Filter where 'var' is FALSE (shorthand)
> dt[var == F]
Error: comparison (1) is possible only for atomic and list types
# Filter where 'var' is FALSE
> dt[var == FALSE]
     var val
1: FALSE   2

This behavior seems pretty undesirable (especially since T and F are so much quicker to type).

Thanks for considering my issue!

[SebKrantz]

Hi, thanks, this has come up before. I can think about depreciating F, but I actually wouldn't like that because it's indeed a backward compatibility issue. Th solution to your code is simply creating F <- FALSEin the global environment. In base R Fis simply a variable exported from the base package, just like collapse exports variables like .FAST_STAT_FUN.

[edrubin]

Totally understand the backwards compatibility issue, but it seems like many more people use F as FALSE than as collapse::F(). Your solution would also work for backwards compatibility issues, e.g., F <- FF.

Either way, thanks for your work on a terrific package—and for your quick reply!

[SebKrantz]

Thanks, I agree with that. I wouldn't rename F to FFbut simply remove it. On the other hand I don't totally dislike the idea that using collapse would compell some peopel to write more secure code to avoid having to create an additional variable F. I'll keep thinking about it.