Skip to content

rgdal problem with wintri projection (OSX, R-3.3.0, rgdal 1.1-9)

9 messages · David Winsemius, Simon Urbanek, Dan Kelley +1 more

Original
Dan Kelley
# 
On OSX, I find that inverse projections yield segmentation faults with the newly-released R (version 3.3.0). This works with both rgdal 1.1-8 and also with 1.1-9 (released yesterday).

Below my signature line is a transcript, within which it is shown that the forward projection from lon=lat=0 yields x=y=0 as expected, but that trying to do the inverse projection with x=y=0 yields a segmentation fault. I?ve put in a sessionInfo() to show the machine type, etc.

I note that aitoff also fails. I?ve not tested the full suite of projections.

NOTE: this test works fine with R-3.2.1.

Dan E. Kelley, Professor  and Graduate Coordinator
Oceanography Department, Dalhousie University
PO BOX 15000
Halifax, NS B3H 4R2
phone:(902)494-1694 fax:(?)-3877 Dan.Kelley at Dal.CA  
http://oceanography.dal.ca/person/Kelley_Dan.html


$ R --no-save < wintri.R

R version 3.3.0 (2016-05-03) -- "Supposedly Educational"
Copyright (C) 2016 The R Foundation for Statistical Computing
Platform: x86_64-apple-darwin13.4.0 (64-bit)

R is free software and comes with ABSOLUTELY NO WARRANTY.
You are welcome to redistribute it under certain conditions.
Type 'license()' or 'licence()' for distribution details.

  Natural language support but running in an English locale

R is a collaborative project with many contributors.
Type 'contributors()' for more information and
'citation()' on how to cite R or R packages in publications.

Type 'demo()' for some demos, 'help()' for on-line help, or
'help.start()' for an HTML browser interface to help.
Type 'q()' to quit R.
R version 3.3.0 (2016-05-03)
Platform: x86_64-apple-darwin13.4.0 (64-bit)
Running under: OS X 10.11.5 (El Capitan)

locale:
[1] en_CA.UTF-8/en_CA.UTF-8/en_CA.UTF-8/C/en_CA.UTF-8/en_CA.UTF-8

attached base packages:
[1] stats     graphics  grDevices utils     datasets  methods   base
Loading required package: sp
rgdal: version: 1.1-9, (SVN revision 617M)
 Geospatial Data Abstraction Library extensions to R successfully loaded
 Loaded GDAL runtime: GDAL 1.11.4, released 2016/01/25
 Path to GDAL shared files: /Library/Frameworks/R.framework/Versions/3.3/Resources/library/rgdal/gdal
 Loaded PROJ.4 runtime: Rel. 4.8.0, 6 March 2012, [PJ_VERSION: 480]
 Path to PROJ.4 shared files: /Library/Frameworks/R.framework/Versions/3.3/Resources/library/rgdal/proj
 Linking to sp version: 1.2-3
lon lat
[1,]   0   0
*** caught segfault ***
address 0x0, cause 'memory not mapped'

Traceback:
 1: .Call("project_inv", as.integer(nc), as.double(xy[, 1]), as.double(xy[,     2]), proj, as.logical(use_ob_tran), PACKAGE = "rgdal")
 2: rgdal::project(xy, proj = proj, inv = TRUE)
An irrecoverable exception occurred. R is aborting now ...
Segmentation fault: 11
David Winsemius
# 
I was using slightly out of date versions of the R packages. With rgdal 1.1-8 and sp 1.2-3 there was no segfault. I thought that the difference might be due to the fact that I'm using a later version of PROJ.4 runtime: Rel. 4.9.2, 08 September 2015, [PJ_VERSION: 492].  However, "updating" to rgdal 1.1-9 now produces the same segfault.
Roger Bivand
# 
David Winsemius <dwinsemius <at> comcast.net> writes:
the newly-released R (version
yesterday).
...
as.double(xy[,     2]), proj,
This appears to affect 1.1-9 on OSX. As rgdal maintainer, obviously I'd like
to resolve this, but have mo access to OSX hardware at all. I'm assuming
that the rgdal version is installed as binary from CRAN (thanks to Simon for
providing the binary package!).

Do we know whether the OSX version matters?

Do we know that the same rgdal version 1.1-9 installed from source suffers
from the same problem; if it does, is 1.1-8 free of the problem (as with the
CRAN binary)?

There may be a user-facing work-around - use spTransform (or rawTransform())
rather than project(), but I cannot check it myself:

xy <- rawTransform("+proj=longlat", "+proj=wintri", n=1L, x=0, y=0)
rawTransform("+proj=wintri", "+proj=longlat", n=1L, x=xy[[1]][1], y=xy[[2]][1])

This is using different functions in the PROJ.4 library.

If anyone could provide gdb or similar output of a backtrace, that would be
very helpful.

Roger
1.1-8 and sp 1.2-3 there was no
using a later version of PROJ.4
"updating" to rgdal 1.1-9 now
Dan Kelley
# 
I apologize for not providing sufficient information in my original posting.  Below I am putting, in markdown notation, the test results with two setups.  I think the key thing is that R 3.3.0 is linking with an old version of proj (4.8.0) whereas R 3.2.1 was linking with a new version (4.9.1).


# Test 1 with R 3.2.1 (works)

    R version 3.2.1 (2015-06-18) -- "World-Famous Astronaut"
    Copyright (C) 2015 The R Foundation for Statistical Computing
    Platform: x86_64-apple-darwin10.8.0 (64-bit)
    
    R is free software and comes with ABSOLUTELY NO WARRANTY.
    You are welcome to redistribute it under certain conditions.
    Type 'license()' or 'licence()' for distribution details.
    
      Natural language support but running in an English locale
    
    R is a collaborative project with many contributors.
    Type 'contributors()' for more information and
    'citation()' on how to cite R or R packages in publications.
    
    Type 'demo()' for some demos, 'help()' for on-line help, or
    'help.start()' for an HTML browser interface to help.
    Type 'q()' to quit R.
    
    > sessionInfo()
    R version 3.2.1 (2015-06-18)
    Platform: x86_64-apple-darwin10.8.0 (64-bit)
    Running under: OS X 10.11.5 (unknown)
    
    locale:
    [1] en_CA.UTF-8/en_CA.UTF-8/en_CA.UTF-8/C/en_CA.UTF-8/en_CA.UTF-8
    
    attached base packages:
    [1] stats     graphics  grDevices utils     datasets  methods   base     
    > library(rgdal)
    Loading required package: sp
    rgdal: version: 0.9-3, (SVN revision 530)
     Geospatial Data Abstraction Library extensions to R successfully loaded
     Loaded GDAL runtime: GDAL 1.11.2, released 2015/02/10
     Path to GDAL shared files: /Library/Frameworks/R.framework/Versions/3.2/Resources/library/rgdal/gdal
     Loaded PROJ.4 runtime: Rel. 4.9.1, 04 March 2015, [PJ_VERSION: 491]
     Path to PROJ.4 shared files: /Library/Frameworks/R.framework/Versions/3.2/Resources/library/rgdal/proj
     Linking to sp version: 1.1-0 
    > lon <- 0
    > lat <- 0
    > proj <- "+proj=wintri"
    > xy <- rgdal::project(cbind(lon, lat), proj=proj)
    > print(xy)
         [,1] [,2]
    [1,]    0    0
    > rgdal::project(xy, proj=proj, inv=TRUE)
         [,1] [,2]
    [1,]    0    0
    > 
    > 




# Test 2 with R 3.3.0 (seg-faults)


    R version 3.3.0 (2016-05-03) -- "Supposedly Educational"
    Copyright (C) 2016 The R Foundation for Statistical Computing
    Platform: x86_64-apple-darwin13.4.0 (64-bit)
    
    R is free software and comes with ABSOLUTELY NO WARRANTY.
    You are welcome to redistribute it under certain conditions.
    Type 'license()' or 'licence()' for distribution details.
    
      Natural language support but running in an English locale
    
    R is a collaborative project with many contributors.
    Type 'contributors()' for more information and
    'citation()' on how to cite R or R packages in publications.
    
    Type 'demo()' for some demos, 'help()' for on-line help, or
    'help.start()' for an HTML browser interface to help.
    Type 'q()' to quit R.
    
    > sessionInfo()
    R version 3.3.0 (2016-05-03)
    Platform: x86_64-apple-darwin13.4.0 (64-bit)
    Running under: OS X 10.11.5 (El Capitan)
    
    locale:
    [1] en_CA.UTF-8/en_CA.UTF-8/en_CA.UTF-8/C/en_CA.UTF-8/en_CA.UTF-8
    
    attached base packages:
    [1] stats     graphics  grDevices utils     datasets  methods   base     
    > library(rgdal)
    Loading required package: sp
    rgdal: version: 1.1-9, (SVN revision 617M)
     Geospatial Data Abstraction Library extensions to R successfully loaded
     Loaded GDAL runtime: GDAL 1.11.4, released 2016/01/25
     Path to GDAL shared files: /Library/Frameworks/R.framework/Versions/3.3/Resources/library/rgdal/gdal
     Loaded PROJ.4 runtime: Rel. 4.8.0, 6 March 2012, [PJ_VERSION: 480]
     Path to PROJ.4 shared files: /Library/Frameworks/R.framework/Versions/3.3/Resources/library/rgdal/proj
     Linking to sp version: 1.2-3 
    > lon <- 0
    > lat <- 0
    > proj <- "+proj=wintri"
    > xy <- rgdal::project(cbind(lon, lat), proj=proj)
    > print(xy)
         lon lat
    [1,]   0   0
    > rgdal::project(xy, proj=proj, inv=TRUE)
    
     *** caught segfault ***
    address 0x0, cause 'memory not mapped'
    
    Traceback:
     1: .Call("project_inv", as.integer(nc), as.double(xy[, 1]), as.double(xy[,     2]), proj, as.logical(use_ob_tran), PACKAGE = "rgdal")
     2: rgdal::project(xy, proj = proj, inv = TRUE)
    An irrecoverable exception occurred. R is aborting now ...
Simon Urbanek
# 
Process 6803 stopped
* thread #1: tid = 0xc72bb10, 0x0000000000000000, queue = 'com.apple.main-thread', stop reason = EXC_BAD_ACCESS (code=1, address=0x0)
    frame #0: 0x0000000000000000
error: memory read failed for 0x0
(lldb) bt
* thread #1: tid = 0xc72bb10, 0x0000000000000000, queue = 'com.apple.main-thread', stop reason = EXC_BAD_ACCESS (code=1, address=0x0)
  * frame #0: 0x0000000000000000
    frame #1: 0x00000001068a2657 rgdal.so`pj_inv(xy=(x = 0, y = 0), P=0x000000010070ac10) + 215 at pj_inv.c:22
    frame #2: 0x0000000106011f2b rgdal.so`project_inv(n=<unavailable>, x=0x0000000105db1d28, y=0x0000000105db1c68, projarg=<unavailable>, ob_tran=<unavailable>) + 251 at projectit.cpp:300
    frame #3: 0x000000010007b827 libR.dylib`do_dotcall(call=0x0000000105b632b0, op=<unavailable>, args=<unavailable>, env=<unavailable>) + 327 at dotcode.c:1251
    frame #4: 0x00000001000a8e4c libR.dylib`Rf_eval(e=0x0000000105b632b0, rho=0x0000000105b7a408) + 988 at eval.c:713

I'll look into the PROJ4 version issue as well and keep you posted.

Cheers,
Simon
On May 11, 2016, at 6:44 AM, Roger Bivand <Roger.Bivand at nhh.no> wrote:

            

      
      
      
    

        

      
      
    

  
  


  


      

    

    

      
      

        


  

        

      Simon Urbanek
    

    

      
      #
    

  


  

      
Thanks, that was the correct hint - updating PROJ4 to 4.9.1 seems to fix the issue. The new binary is now up - until all mirrors sync up please use the main Mac server:
install.packages("rgdal",,"http://r.research.att.com")

Roger, can you, please, add a corresponding regression test to rgdal so a failure like the one below would be picked up on check?

Thanks,
Simon

        
On May 11, 2016, at 7:25 AM, Daniel Kelley <Dan.Kelley at Dal.Ca> wrote:

        

            

      
      
      
    

        

      
      
    

  
  


  


      

    

    

      
      

        


  

        

      Dan Kelley
    

    

      
      #
    

  


  

      
This is fantastic ? thanks, Simon and Roger! Users of the ?oce? package (which uses rgdal for map-projection calculations) will be very grateful.

Dan.

Dan E. Kelley, Professor  and Graduate Coordinator
Oceanography Department, Dalhousie University
PO BOX 15000
Halifax, NS B3H 4R2
phone:(902)494-1694 fax:(?)-3877 Dan.Kelley at Dal.CA  
http://oceanography.dal.ca/person/Kelley_Dan.html

  
  


  


      

    

    

      
      

        


  

        

      Roger Bivand
    

    

      
      #
    

  


  

      
On Wed, 11 May 2016, Simon Urbanek wrote:

        

            

      
      
      
    

        
Yes, of course, and thank you for resolving this so fast. The next release 
will also have extra protection contributed by Barry Rowlingson (still 
testing).

Roger

            

      
      
      
    

        

      
      
    

            

      
      
      
    

        

  
    

      
      
    

  


  


      

    

    

      
      

        


  

        

      Roger Bivand
    

    

      
      #
    

  


  

      
On Wed, 11 May 2016, Roger Bivand wrote:

        

            

      
      
      
    

        
rgdal_1.1-10 submitted to CRAN with additional tests which reproduce the 
issue (as far as I can tell without being able to check directly), and 
with a fix by Barry Rowlingson for another problem identified while trying 
to add the tests. In addition, project() now has an added argument 
legacy=, which if TRUE will use PROJ.4 pj_fwd() and pj_inv(), or if FALSE 
will use pj_transform() (on Windows 32-bit it will always use 
pj_transform()).

The sources are committed to R-Forge.

Anyone able to test this version is very welcome to try it out.

Roger