R devel - Mar 2002 - rare bad bug in sys.function() {or match.arg()} (PR#1409)

I found this tracing a bug when experimenting with a new sort()
function using match.arg().

It was triggered because mosaicplot.default(.) has an argument
called `sort' and calls itself the sort() function  in which I
was using match.arg()...

Here is (input for) a small clean example :

####  bad  match.arg() // sys.function() bug :
####  MM, 22.3.2002

callme <- function(a = 1, mm = c("Abc", "Bde")){
    mm <- match.arg(mm)
    cat("mm = "); str(mm)
    invisible()
}


## These are as desired
callme()
callme(mm="B")

mycaller <- function(x = 1, callme = pi) {
    callme(x)
}

mycaller()
##-> mm = NULL

## but should give "Abc !!

-------

Explanation :

 In match.arg(), there's
     formal.args <- formals(sys.function(sys.parent()))
 and you can check that  sys.function(sys.parent())
 gives the number pi instead of the the callme function !! :

debug(match.arg)
mycaller()
.....

Martin Maechler <maechler@stat.math.ethz.ch>
http://stat.ethz.ch/~maechler/
Seminar fuer Statistik, ETH-Zentrum  LEO C16	Leonhardstr. 27
ETH (Federal Inst. Technology)	8092 Zurich	SWITZERLAND
phone: x-41-1-632-3408		fax: ...-1228			<><

-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-
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
_._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._

maechler@stat.math.ethz.ch writes:
> I found this tracing a bug when experimenting with a new sort()
> function using match.arg().
> 
> It was triggered because mosaicplot.default(.) has an argument
> called `sort' and calls itself the sort() function  in which I
> was using match.arg()...
> 
> Here is (input for) a small clean example :
> 
> ####  bad  match.arg() // sys.function() bug :
> ####  MM, 22.3.2002
> 
> callme <- function(a = 1, mm = c("Abc", "Bde")){
>     mm <- match.arg(mm)
>     cat("mm = "); str(mm)
>     invisible()
> }
> 
> 
> ## These are as desired
> callme()
> callme(mm="B")
> 
> mycaller <- function(x = 1, callme = pi) {
>     callme(x)
> }
> 
> mycaller()
> ##-> mm = NULL
> 
> ## but should give "Abc !!
> 
> -------
> 
> Explanation :
> 
>  In match.arg(), there's
>      formal.args <- formals(sys.function(sys.parent()))
>  and you can check that  sys.function(sys.parent())
>  gives the number pi instead of the the callme function !! :
> 
> debug(match.arg)
> mycaller()
> .....
This looks suspiciously like the same issue as PR#921. Did someone
have plans to change the context stack handling so that we could have
a "real" sys.function()? To quote my own followup to PR#921:

"The problem is that we only actually store the call in the context
structure, not the function, so sys.function has to grab the call,
take the 1st element, and reevaluate it in the parent frame."



-- 
   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
_._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._

>>>>> "PD" == Peter Dalgaard BSA
<p.dalgaard@biostat.ku.dk> writes:
    PD> maechler@stat.math.ethz.ch writes:
    >> I found this tracing a bug when experimenting with a new
    >> sort() function using match.arg().
    >> 
    >> It was triggered because mosaicplot.default(.) has an
    >> argument called `sort' and calls itself the sort()
    >> function in which I was using match.arg()...
    >> 
    >> Here is (input for) a small clean example :
    >> 
    >> #### bad match.arg() // sys.function() bug : #### MM,
    >> 22.3.2002
    >> 
    >> callme <- function(a = 1, mm = c("Abc",
"Bde")){ mm <-
    >> match.arg(mm) cat("mm = "); str(mm) invisible() }
    >> 
    >> 
    >> ## These are as desired callme() callme(mm="B")
    >> 
    >> mycaller <- function(x = 1, callme = pi) { callme(x) }
    >> 
    >> mycaller() ##-> mm = NULL
    >> 
    >> ## but should give "Abc !!
    >> 
    >> -------
    >> 
    >> Explanation :
    >> 
    >> In match.arg(), there's formal.args <-
    >> formals(sys.function(sys.parent())) and you can check
    >> that sys.function(sys.parent()) gives the number pi
    >> instead of the the callme function !! :
    >> 
    >> debug(match.arg) mycaller() .....

    PD> This looks suspiciously like the same issue as
    PD> PR#921. Did someone have plans to change the context
    PD> stack handling so that we could have a "real"
    PD> sys.function()? To quote my own followup to PR#921:

    PD> "The problem is that we only actually store the call in
    PD> the context structure, not the function, so sys.function
    PD> has to grab the call, take the 1st element, and
    PD> reevaluate it in the parent frame."

hmm, bad news.  
but couldn't it try to do the equivalent of
    get(name, mode = "function")
instead of simply
    get(name)
?

[I think this *must* be fixed, at least we'd have to program
  around it in match.arg()]

Martin

-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-
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
_._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._