R devel - Jan 1998 - R-beta: sequences

A non-text attachment was scrubbed...
Name: not available
Type: text
Size: 1659 bytes
Desc: not available
Url :
https://stat.ethz.ch/pipermail/r-help/attachments/19980107/3a4902ef/attachment.pl

Jim Lindsey <jlindsey at luc.ac.be> writes:
> 
>   1. The change of sequences from reals to integers has created a number
> of problems. I already mentioned this earlier on the list but here is
> another example:
> contour(1:10,1:10,z)
> where z is an appropriate matrix, does not work. After searching quite
> a while, I discovered in the C code that contour checks for reals and
> does not allow integers. Someone is going to have through all the C
> code and change checks for reals to numeric anywhere where sequences
> might be used.
- or sprinkle as.double() all over the R code...
 
>   2. I am still getting systematic crashes when nlm is used heavily
> (since at least last May when I first told it on the list, so this is
> not new to the latest versions).  When nlm is working on a big problem
> intensively for say 5-10 minutes (Pentium 133 with Linux Red Hat 4.1),
> the X server often crashes. If I try to paste something into emacs,
> emacs alone will often crash, sometimes when that is the only thing I
> am doing, but more often when I adjust a window or click in a
> slider. I can usually, but not always, restart X but it is bound to
> happen again so I usually need to reboot. Sometimes it freezes the
> whole computer and I have to shut it off. It seems to occur mostly (if
> not always) when (log) gamma is being used in the likelihood function
> being optimized. But it is not reproducible so I cannot give an
> example.
This sounds like CPU overheating more than anything else (fan still
running?). Note that in contrast to certain other OS's, Linux doesn't
run a busy wait loop when the system is idle, so CPU-intensive tasks
result in an abrupt rise in temperature. Linux should protect
application programs from crashing other parts of the system, so this
is hardly a fault in R.

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

-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-
r-help 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-help-request at
stat.math.ethz.ch
_._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._

[diverted from R-help to R-devel]
>>>>> "Jim" == Jim Lindsey <jlindsey@luc.ac.be>
writes:
    Jim>   1. The change of sequences from reals to integers has created a
    Jim> number of problems. I already mentioned this earlier on the list
    Jim> but here is another example: contour(1:10,1:10,z) where z is an
    Jim> appropriate matrix, does not work. After searching quite a while,
    Jim> I discovered in the C code that contour checks for reals and does
    Jim> not allow integers. Someone is going to have through all the C
    Jim> code and change checks for reals to numeric anywhere where
    Jim> sequences might be used.

Sequences are not the only integer variables in R, even though probably the
most frequent ones.

I wonder, too what to do.
There is the  asReal(.) macro  which could be used instead of REAL(.),
but I wonder if this is the correct approach.

Ross ?

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