Skip to content

[R-pkg-devel] R CMD build segfault (r88038)

4 messages · Dirk Eddelbuettel, Josiah Parry

#
In trying to address new WARNINGs on R-devel I'm trying to rebuild my
package environment on R 4.6 (r88038) so I can try and remediate. However,
I'm encountering a segfault when doing so which doesn't occur for R 4.5.0
alpha (2025-03-18 r87991).

Is there a known change to building using R CMD build that we have to
address?

rextendr ? R CMD build .
* checking for file ?./DESCRIPTION? ... OK
* preparing ?rextendr?:
* checking DESCRIPTION meta-information ... OK
* installing the package to build vignettes
      -----------------------------------
* installing *source* package ?rextendr? ...
** this is package ?rextendr? version ?0.3.1.9001?
** using staged installation
** R
** inst
** byte-compile and prepare package for lazy loading
sh: line 1: 20659 Segmentation fault: 11  R_TESTS=
'/Library/Frameworks/R.framework/Versions/4.6-arm64/Resources/bin/R'
--no-save --no-restore --no-echo 2>&1 <
'/var/folders/wd/xq999jjj3bx2w8cpg7lkfxlm0000gn/T//RtmpkyfWsl/file50adcb72e68'
#
On 24 March 2025 at 09:10, Josiah Parry wrote:
| In trying to address new WARNINGs on R-devel I'm trying to rebuild my
| package environment on R 4.6 (r88038) so I can try and remediate. However,
| I'm encountering a segfault when doing so which doesn't occur for R 4.5.0
| alpha (2025-03-18 r87991).
| 
| Is there a known change to building using R CMD build that we have to
| address?
| 
| rextendr ? R CMD build .
| * checking for file ?./DESCRIPTION? ... OK
| * preparing ?rextendr?:
| * checking DESCRIPTION meta-information ... OK
| * installing the package to build vignettes
|       -----------------------------------
| * installing *source* package ?rextendr? ...
| ** this is package ?rextendr? version ?0.3.1.9001?
| ** using staged installation
| ** R
| ** inst
| ** byte-compile and prepare package for lazy loading
| sh: line 1: 20659 Segmentation fault: 11  R_TESTS=
| '/Library/Frameworks/R.framework/Versions/4.6-arm64/Resources/bin/R'
| --no-save --no-restore --no-echo 2>&1 <
| '/var/folders/wd/xq999jjj3bx2w8cpg7lkfxlm0000gn/T//RtmpkyfWsl/file50adcb72e68'

It's been a while since I have seen such a segfault, but I have encountered
plenty in both the earlier days of Rcpp, or when working with for example the
geospatial satck. And it almost always traced back to different (binary,
i.e. 'native' code containing) packages requiring a rebuild.  While I haven't
seen it lately, my suggestion would still be to see if it persists if you
rebuild everything touched by that package / if it happens for subsets of
these packages. Subsetting to the other native packages may suffice. From a
quick check (see below) that is still 16 packages.

Dirk
[1] "brio"       "callr"      "cli"        "desc"       "dplyr"     
 [6] "glue"       "jsonlite"   "pkgbuild"   "processx"   "purrr"     
[11] "rlang"      "rprojroot"  "stringi"    "tibble"     "vctrs"     
[16] "withr"      "R6"         "utils"      "generics"   "lifecycle" 
[21] "magrittr"   "methods"    "pillar"     "tidyselect" "ps"        
[26] "tools"      "stats"      "fansi"      "pkgconfig"  "graphics"  
[31] "grDevices"  "utf8"
[1] "brio"       "cli"        "dplyr"      "glue"       "jsonlite"  
 [6] "processx"   "purrr"      "rlang"      "stringi"    "tibble"    
[11] "vctrs"      "magrittr"   "tidyselect" "ps"         "fansi"     
[16] "utf8"

  
    
#
?

Josiah Parry reacted via Gmail
<https://www.google.com/gmail/about/?utm_source=gmail-in-product&utm_medium=et&utm_campaign=emojireactionemail#app>
On Mon, Mar 24, 2025 at 9:48?AM Dirk Eddelbuettel <edd at debian.org> wrote:

            
#
Thank you! I suspect it is because I installed stringi from a binary
created on r-universe and it may not have been compatible with my machine
or version of R-devel.
On Mon, Mar 24, 2025 at 9:48?AM Dirk Eddelbuettel <edd at debian.org> wrote: