search

noamross / zero-dependency-problems

Real-world code problems in R, without dependencies
79 stars 25 forks source link

Comments on "Common errors in R" #19

Open noamross opened 9 years ago

noamross commented 9 years ago

Comments on https://github.com/noamross/zero-dependency-problems/blob/master/misc/stack-overflow-common-r-errors.md

dpastoor commented 9 years ago

nice analysis!

regarding

"Error in if" generally means the logical statement in "if (XXX) { ..." is not yielding a logical value. Most of these have missing value where TRUE/FALSE needed, meaning that the variable in the if statement can't be found.

There are some good 'lessons' to come out of digging into parameter 'truthiness'. The most common of which is to explain first why if(param ==TRUE) is unnecessary, and second once they understand inherently that if statements need boolean values, to introduce how to handle statments that only sometimes return booleans.

if(param) ...

vs

 if(isTRUE(param)) {}

I find using all.equal as an example for when isTRUE is appropriate.

The all equal can also lead to a nice aside about floating point numbers!

0.3_3 == 0.9 [1] FALSE all.equal(0.3_3, 0.9) [1] TRUE

Though frankly I wouldn't consider these examples suitable for a beginners workshop (more important things to cover in limited time imo), but are good reference material that should be easily accessible when they are ready to keep exploring.

revodavid commented 9 years ago

Very nice analysis! Thanks for the discussion about all the various types of errors, though you seem to have skipped #10, the "missing value where" error type, which I presume is mainly the result of missing values in data where it wasn't expected.

EDIT: On reading closer you did mention it -- those "missing value where" errors were actually bad calls to if.

mxxdxxx commented 9 years ago

Haha nice and pretty useful, thanks! Personally I would have expected "more columns than column names" to be in the list. But maybe it's just a prominent error which is not asked that often.

hdarabi commented 9 years ago

Great and funny analysis. Thanks!

benmarwick commented 7 years ago

What's your preference for citing this? Has it been published somewhere?

noamross commented 7 years ago

@benmarwick It hasn't been published elsewhere. I guess it makes sense to just cite it as one would a blog post somewhere.