Why am I referred to the manual page for "Arithmetic" when I type `?mgcv-faq` without `library(mgcv)`?

Question

Well I spot this when I actually made a mistake. I want to get the manual page of mgcv.FAQ, but I

• forgot to do library(mgcv);
• wrongly put ?mgcv-faq.

But, R strangely directs me to the doc page as if I have done ?Arithmetic.

What is going on? After I do library(mgcv), putting ?mgcv-faq now gives an error:

#Error in eval(argExpr, envir) : object 'mgcv' not found
#Error in .signatureFromCall(fdef, expr, envir, doEval) : 
#  error in trying to evaluate the expression for argument ‘e1’ (mgcv)

Can anyone explain this behavior?

---

sessionInfo() before library(mgcv):

R version 3.4.4 (2018-03-15)
Platform: x86_64-pc-linux-gnu (64-bit)
Running under: Ubuntu 16.04.5 LTS

Matrix products: default
BLAS/LAPACK: /usr/lib/libopenblas.so

locale:
 [1] LC_CTYPE=en_GB.UTF-8       LC_NUMERIC=C              
 [3] LC_TIME=en_GB.UTF-8        LC_COLLATE=en_GB.UTF-8    
 [5] LC_MONETARY=en_GB.UTF-8    LC_MESSAGES=en_GB.UTF-8   
 [7] LC_PAPER=en_GB.UTF-8       LC_NAME=C                 
 [9] LC_ADDRESS=C               LC_TELEPHONE=C            
[11] LC_MEASUREMENT=en_GB.UTF-8 LC_IDENTIFICATION=C      

attached base packages:
[1] stats     graphics  grDevices utils     datasets  methods   base     

loaded via a namespace (and not attached):
[1] compiler_3.4.4

attached base packages:
[1] stats     graphics  grDevices utils     datasets  methods   base    

sessionInfo() after library(mgcv):

## only shows what is different from the above

other attached packages:
[1] mgcv_1.8-24  nlme_3.1-137

loaded via a namespace (and not attached):
[1] compiler_3.4.4  Matrix_1.2-14   tools_3.4.4     grid_3.4.4     
[5] lattice_0.20-35

I don't have Rstudio, so I run R in the terminal. And the manual page is displayed in the terminal.

Answer 1

You can reproduce this with:

expr <- quote(mgcv-faq)
utils:::.helpForCall(expr, parent.frame())

.helpForCall gets called by ? internally with this input.

Now, if the expression does not contain :: or ::: calls, .helpForCall extracts the symbol to look up by doing

expr[[1L]]
#`-`

without checking the length of the expression.

The - operator is first in the expression because the expression is parsed to function syntax.

Why does this happen? Because ? should be able to handle :: and ::: in its input and nobody thought of handling a user error such as yours.