Skip to content

Status Rmetrics - New Version

3 messages · Diethelm Wuertz, Spencer Graves, Dirk Eddelbuettel

Original
Diethelm Wuertz
# 
Problems to be solved with Rmetrics for R-2.2.0
    
    I'm aware of the following problems which will be very
    likely be patched for the Rmetrics Version for R 2.2.0
    
    1.  Changing the compiler from g77 to gfortran
        under Linux created some severe problems and errors.
        STATUS: already solved, fixed with next Rmetrics Version
        
    2.  Does the windows version also compile under gfortran?
        STATUS: yes - I created my own mingw gfortran suite
        
    3.  fCalendar and chron create conflicts and deliver
        wrong results when both are loaded at the same time.
        STATUS: not yet solved - maybe fixed with next version
        
    4.  Markowitz approach from fPortfolio has some bugs ...
        STATUS They will be fixed with the next version of Rmetrics
        
    5.  listFinCenter got a problem with version 2.1.
        STATUS: already solved, fixed with next Rmetrics Version
        
    6.  timeDate got a problem under Windows with version 2.1.
        STATUS: already solved, fixed with next Rmetrics Version
        
    *** The next updated Rmetrics version for R 2.2.0 will become ***
    *** available as soon as possbile, likely within October.     ***
    
    Did you know?
    
        Rmetrics comes with 1399 S functions and 95945 lines of R-code!
        RUnit testing is under progress to improve software quality.
    
    I apologize for any inconveniances caused by bugs and
    software updates.
    
    Diethelm Wuertz
Spencer Graves
# 
Hi, Diethelm:

	  Thank you for all your hard work in assembling this package and 
making it available to the R community.

	  Best Wishes,
	  Spencer Graves
Diethelm Wuertz wrote:

            

      
      
      
    

        

      
      
    

        

  
    

      
      
    

  


  


      

    

    

      
      

        


  

        

      Dirk Eddelbuettel
    

    

      
      #
    

  


  

      
Hi Diethelm,

Thanks for the status update -- much appreciated.

        
On 13 October 2005 at 15:49, Diethelm Wuertz wrote:

        
|     Problems to be solved with Rmetrics for R-2.2.0
|     
|     I'm aware of the following problems which will be very
|     likely be patched for the Rmetrics Version for R 2.2.0
|     
|     1.  Changing the compiler from g77 to gfortran
|         under Linux created some severe problems and errors.
|         STATUS: already solved, fixed with next Rmetrics Version

As far as I know (which may not mean much as I don;t follow compiler issues
all that closely) that GCC upstream does not yet recommend to use gfortran. 
I think Brian Ripley stated about the same on one of the R lists.  

As another data point, Debian switched its default compiler to gcc 4.0.* for
every language but Fortran where g77 from gcc 3.4.* is still used (and 3.4
and 4.0 are said to produce interchangeable object code).  

So it may be too soon to start fixing wholesale for gfortran. Just my $0.02.

The rest looks good -- looking forward to the next Rmetrics release!

Regards, Dirk