[Rd] Strange parse behavior (PR#8815)

[murdoch at stats.uwo.ca]

On 4/27/2006 3:06 PM, Thomas Friedrichsmeier wrote:
> Hi,
> 
> I discovered the following strange behavior in parse () on syntactically 
> incorrect statements. This is on R 2.3.0 (linux). It may or may not have been 
> present earlier than 2.3.0, but I only discovered it recently. I can see no 
> mention of it in the NEWS file from trunk.
> 
> Consider these statements (the output I get is shown commented below each 
> statement):
> 
> bad.syntax <- ")\nprint (\"hello\")\nprint (\"world\")"
> 
> try (parse (text=bad.syntax, n=-1))	# OK
> # Error in parse(file, n, text, prompt) : syntax error in ")"
> try (parse (text=bad.syntax, n=1))	# OK
> # Error in parse(file, n, text, prompt) : syntax error in ")"
> try (parse (text=bad.syntax, n=2))	# No error!
> # NULL
> try (parse (text=bad.syntax, n=3))	# No error!
> # NULL
> try (parse (text=bad.syntax, n=4))	# Misleading message
> # Error in parse(file, n, text, prompt) : syntax error in:
> # "print ("hello")
> # print ("world")"
> 
> Probably there are not too many use cases of parse (n=x) with x not either -1 
> or 1, so it should not be a grave problem, but it just doesn't look right.

I see the same results as you, in versions back to 1.9.1.  I agree that 
it looks like a bug, but not a new one.  I've sent this reply to the 
bugs list so it gets recorded.  I won't have time to look into it soon.

Duncan Murdoch