Easy Made Mistakes when Writing R code

Things under legendu.net/outdated are outdated technologies that the author does not plan to update any more. Please look for better alternatives.

Being flexible is a two-side sword to R. While it make it convenient and productive to use R, it is also very easy to make mistakes when writing R code. The following is list of mistakes that happens all the time.

1. Miss passed arguments
   This is often due the to "dots" argument (...). For example, I used to use VB a lot, so when I use the function seq, I wrote code like this

   seq(1,10,step=2)

Another example is

cat(data,file_name)

You thought you write data into the file file_name, however you just print data and file_name to the console. The annoying things about dots argument is that it accept any kind of arguments. So whatever you pass to a function accepting a dots argument, no warning or error about argument passing will be shown. The dots argument opens the Pandor's box.

1. Global variables
   While you can assign values to global variables using =, <- and -> inside a user-defined function, the change (of the global variable) is only in effect in the duration of the function. When the function ends, the change is no longer in effect. In another words, the global variable remains unchanged outside the function. To make the change in effect outside the function, you must use <<- and ->> to assign values to the global variable.

2. Priority of operators The colon (:) operator precede over arithmatic operators such as +, -, *, / and so on. However, [] precede over the colon operator. When you write code 1:n-1, you thought you get a vector from 1 to n-1 while you actually get a vector from 0 to n-1.

3. Miss used functions The function length is used for querying length of vectors not the length of characters. For the later purpose, use the function nchar.

4. Use && or || on vectors
   For vector operation, you should use & and |.

5. Use vectors or missing value as conditions
   For if and while statement, the condition must be a non-missing scaler (a vector of length 1).

6. Use *apply functions to operate on rows of data frames. A row of data frame is still a data frame (with 1 row). R is evil in the sence that it coerce data type wheneve necessary without noticing users. ...