R devel - Mar 2000 - using function names as parameters (PR#469)

When using a function name as parameter, here date, the following happends: 

R> foo <- function(date=date()) cat(date, "\n")
R> foo()
Error in cat(date, "\n") : recursive default argument reference

R> foo("Good Morning!")
Good Morning!

works.

Torsten

--please do not edit the information below--

Version:
 platform = i686-unknown-linux
 arch = i686
 os = linux
 system = i686, linux
 status = 
 major = 1
 minor = 0.0
 year = 2000
 month = February
 day = 29
 language = R

Search Path:
 .GlobalEnv, Autoloads, package:base

-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-
r-devel mailing list -- Read http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html
Send "info", "help", or "[un]subscribe"
(in the "body", not the subject !)  To:
r-devel-request@stat.math.ethz.ch
_._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._

hothorn@ci.tuwien.ac.at writes:
> When using a function name as parameter, here date, the following happends:
> 
> R> foo <- function(date=date()) cat(date, "\n")
> R> foo()
> Error in cat(date, "\n") : recursive default argument reference
This is not a bug. Default arguments are evaluated in the frame of the
called function, so when the date variable is requested and no actual
argument was supplied, R tries to evaluate the default expression, but
that involved the variable "date" again. If you pass an argument, the
default expression is never evaluated and the problem does not arise.

Similarly a definition like

function(x=y, y=x) x + y

is asking for trouble if both x and y are left unspecified on the call.

[This has nothing to do with passing functions as parameters, by the
way] 
-- 
   O__  ---- Peter Dalgaard             Blegdamsvej 3  
  c/ /'_ --- Dept. of Biostatistics     2200 Cph. N   
 (*) \(*) -- University of Copenhagen   Denmark      Ph: (+45) 35327918
~~~~~~~~~~ - (p.dalgaard@biostat.ku.dk)             FAX: (+45) 35327907
-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-
r-devel mailing list -- Read http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html
Send "info", "help", or "[un]subscribe"
(in the "body", not the subject !)  To:
r-devel-request@stat.math.ethz.ch
_._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._

>>>>> On 03 Mar 2000 12:55:21 +0100,
>>>>> Peter Dalgaard BSA (PDB) wrote:
PDB> hothorn@ci.tuwien.ac.at writes:
>> When using a function name as parameter, here date, the following
happends:
>> 
R> foo <- function(date=date()) cat(date, "\n")
R> foo()
>> Error in cat(date, "\n") : recursive default argument
reference
PDB> This is not a bug. Default arguments are evaluated in the frame of the
PDB> called function, so when the date variable is requested and no actual
PDB> argument was supplied, R tries to evaluate the default expression, but
PDB> that involved the variable "date" again. If you pass an
argument, the
PDB> default expression is never evaluated and the problem does not arise.

PDB> Similarly a definition like

PDB> function(x=y, y=x) x + y

PDB> is asking for trouble if both x and y are left unspecified on the call.

PDB> [This has nothing to do with passing functions as parameters, by the
PDB> way] 

[Taken to r-core]

Hmm, I asked Torsten to mail the bug report ... I first thought along
your lines and I would agree if I pass ``date'' as default, but we
have ``date()'' as default argument which is the return value of the
function, no?

IMO evaluating the default expression comes down to evaluating
``date <- date()'' which is perfectly legal ...

BTW, BB can do it:

S> foo <- function(date=date()) cat(date, "\n")
S> foo()
Fri Mar  3 09:49:20 MET 2000 
Warning messages:
  looking for function "date", ignored local non-function in: foo()


.f



-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-
r-devel mailing list -- Read http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html
Send "info", "help", or "[un]subscribe"
(in the "body", not the subject !)  To:
r-devel-request@stat.math.ethz.ch
_._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._

Friedrich Leisch <Friedrich.Leisch@ci.tuwien.ac.at> writes:
> Hmm, I asked Torsten to mail the bug report ... I first thought along
> your lines and I would agree if I pass ``date'' as default, but we
> have ``date()'' as default argument which is the return value of
the
> function, no?
> 
> IMO evaluating the default expression comes down to evaluating
> ``date <- date()'' which is perfectly legal ...
> 
> BTW, BB can do it:
> 
> S> foo <- function(date=date()) cat(date, "\n")
> S> foo()
> Fri Mar  3 09:49:20 MET 2000 
> Warning messages:
>   looking for function "date", ignored local non-function in:
foo()
However,
> foo <- function(date=date) date             
> foo() 
Error in foo(): Recursive occurrence of default argument "date"
Dumped

So S-plus only works by assuming that the local variable is not a
function (*before* evaluating its default)! R is being cleaner and
more consistent..

-- 
   O__  ---- Peter Dalgaard             Blegdamsvej 3  
  c/ /'_ --- Dept. of Biostatistics     2200 Cph. N   
 (*) \(*) -- University of Copenhagen   Denmark      Ph: (+45) 35327918
~~~~~~~~~~ - (p.dalgaard@biostat.ku.dk)             FAX: (+45) 35327907

-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-
r-devel mailing list -- Read http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html
Send "info", "help", or "[un]subscribe"
(in the "body", not the subject !)  To:
r-devel-request@stat.math.ethz.ch
_._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._