Skip to content

make check fails for R 2.3.0 (PR#8343)

4 messages · Brian Ripley, Martin Maechler, Arne Henningsen

Original
Brian Ripley
# 
This message is in MIME format.  The first part should be readable text,
  while the remaining parts are likely unreadable without MIME-aware tools.

--27464147-1978628741-1132653680=:25449
Content-Type: TEXT/PLAIN; CHARSET=ISO-8859-1; format=flowed
Content-Transfer-Encoding: QUOTED-PRINTABLE
Content-ID: <Pine.LNX.4.61.0511221102121.26494 at gannet.stats>
On Tue, 22 Nov 2005 ahenningsen at email.uni-kiel.de wrote:

            

      
      
      
    

        
Please use the svn revision for unreleased versions of R.  There is no 'R=
=20
2.3.0' and the version changes several times a day.  But we do expect=20
'make check' to have been run successfully on each of those versions.

            

      
      
      
    

        
You are reporting as a bug in R a problem on your own system in an=20
unreleased ('unstable') version of R.  Since it is unstable and=20
unreleased, such things are by definition not bugs in R.

Others are not seeing this, so we cannot do anything about the problems=20
seen on your system.  This is not at all a new test, and although random=20
it is run with set.seed(1). I can reproduce the result in the output file=
=20
(on my systems) exactly by

            

      
      
      
    

        
Please see if you can debug it on your own system.  (My guess would be=20
that it only occurs as part of the test file.)

            

      
      
      
    

        
es'

            

      
      
      
    

        
s'

            

      
      
      
    

        
s'

            

      
      
      
    

        
--=20
Brian D. Ripley,                  ripley at stats.ox.ac.uk
Professor of Applied Statistics,  http://www.stats.ox.ac.uk/~ripley/
University of Oxford,             Tel:  +44 1865 272861 (self)
1 South Parks Road,                     +44 1865 272866 (PA)
Oxford OX1 3TG, UK                Fax:  +44 1865 272595
--27464147-1978628741-1132653680=:25449--

  
  


  


    

    

      
      

        


  

        

      Arne Henningsen
    

    

      
      #
    

  


  

      
On Tuesday 22 November 2005 12:55, ripley at stats.ox.ac.uk wrote:

            

      
      
      
    

        
OK. I have done this, but the same error still occurs.

            

      
      
      
    

        
I used this version to check my R packages because the packages on CRAN are 
checked by R-devel, too.

            

      
      
      
    

        
Sorry, I did not know this. I thought that my report could help you. 
The next time when I will find an error in R-devel I won't report it.

            

      
      
      
    

        
Yes, that's exactly the case. If you want any further information please don't 
hesitate to contact me. Otherwise I won't bother you anymore with this issue.

Best,
Arne

            

      
      
      
    

        

      
      
    

        

  
    

      
      
    

  


  


      

    

    

      
      

        


  

        

      Martin Maechler
    

    

      
      #
    

  


  

          

      
      
      
    

        
.....

    >> You are reporting as a bug in R a problem on your own system in an=20
    >> unreleased ('unstable') version of R.  

    Arne> I used this version to check my R packages because the
    Arne> packages on CRAN are checked by R-devel, too.

    >> Since it is unstable and
    >> unreleased, such things are by definition not bugs in R.

    Arne> Sorry, I did not know this. I thought that my report could help you. 
    Arne> The next time when I will find an error in R-devel I won't report it.

No; please do "report" the problem, which may be useful for
development, but please do *NOT* use the bug repository, and
probably don't assume it's a bug in R, unless you have quite a
bit experience about R bugs and non-bugs.

Instead, just send e-mail to R-devel and explain,
and you may actually helping R development, particularly if you
are willing to investigate some details that we ma ask you
about.


    >> Others are not seeing this, so we cannot do anything
    >> about the problems=20 seen on your system.  This is not
    >> at all a new test, and although random=20 it is run with
    >> set.seed(1). I can reproduce the result in the output
    >> file= =20 (on my systems) exactly by
    >> 
    >> > set.seed(1)
    >> > hist(replicate(100, mean(rexp(10))))
    >> 
    >> Please see if you can debug it on your own system.  (My guess would be=20
    >> that it only occurs as part of the test file.)

    Arne> Yes, that's exactly the case. If you want any further
    Arne> information please don't hesitate to contact
    Arne> me. Otherwise I won't bother you anymore with this
    Arne> issue.

Too bad.
It might have been interesting to see what

 set.seed(1)
 replicate(100, mean(rexp(10)))

or also

 set.seed(1)
 hist(replicate(100, mean(rexp(10))))
 traceback()
 ##^^^^^^^^^

gives on your R-devel installation.
That's why Brian Ripley helped you by mentioning 'set.seed(1)'.

Regards,
Martin Maechler, ETH Zurich

  
  


  


      

    

    

      
      

        


  

        

      Arne Henningsen
    

    

      
      #
    

  


  

      
On Tuesday 22 November 2005 17:34, Martin Maechler wrote:

            

      
      
      
    

        
It is exactly as Brian Ripley said: If I just start R and execute
   set.seed(1)
   hist(replicate(100, mean(rexp(10))))
everything works well. However, if I run "make check" the error occurs.
Now I have tried something inbetween: I started R and source()d the file that 
caused the error. The error occured again:
R> sink("base-Ex.Rout")
R> source("tests/Examples/base-Ex.R",echo=TRUE)
Warnung in gamma(x) :NaNs wurden erzeugt
Warnung in gamma(x) :NaNs wurden erzeugt
Warnung in gamma(x) :NaNs wurden erzeugt
Fehler in assign("y", 2, env = e) : kann keine Bindungen zu einer 
abgeschlossenen Umgebung hinzuf?gen
Fehler in assign("x", 2, env = e) : kann den Wert einer festgestellten Bindung 
nicht ?ndern
Warnung in cbind(1, 1:7, diag(3)) :number of rows of result
        is not a multiple of vector length (arg 1)
Warnung in cbind(1, 0, matrix(1, nrow = 0, ncol = 4)) :
         number of rows of result
        is not a multiple of vector length (arg 1)
Warnung in withCallingHandlers({ :A
Warnung in data.matrix(DF) :Klasseninformation f?r eine oder mehrere Spalten 
verloren
Garbage collection 25 = 17+1+7 (level 2) ...
221048 cons cells free (47%)
10.2 Mbytes of heap free (90%)
Warnung in sqrt(x) :NaNs wurden erzeugt
Warnung in sqrt(x) :NaNs wurden erzeugt
Warnung in sin(Inf) :NaNs wurden erzeugt
Warnung in cos(Inf) :NaNs wurden erzeugt
Warnung in tan(Inf) :NaNs wurden erzeugt
Fehler in hist.default(replicate(100, mean(rexp(10)))) :
        invalid number of 'breaks'


Then I typed
R> traceback()
6: stop("invalid number of 'breaks'")
5: hist.default(replicate(100, mean(rexp(10))))
4: hist(replicate(100, mean(rexp(10))))
3: eval.with.vis(expr, envir, enclos)
2: eval.with.vis(ei, envir)
1: source("tests/Examples/base-Ex.R")

After sourcing this file it is impossible to use hist() because always the 
same error message occurs:
R> hist(rnorm(55))
Fehler in hist.default(rnorm(55)) : invalid number of 'breaks'
R> traceback()
3: stop("invalid number of 'breaks'")
2: hist.default(rnorm(55))
1: hist(rnorm(55))
R> hist(c(1,2,2))
Fehler in hist.default(c(1, 2, 2)) : invalid number of 'breaks'
R> traceback()
3: stop("invalid number of 'breaks'")
2: hist.default(c(1, 2, 2))
1: hist(c(1, 2, 2))

This seems to be independent from argument x (the values).

Cheers,
Arne