[Rd] R-devel (87913) compilation fails at "make vignettes"; R-4-4 completes and fails "make check"

[Avraham Adler]

Thank you very much, Sebastian. Your suggestion about checking the
FAIL solved the problem. It reminded me that GCC 12+ has a bug
regarding AVX2 alignments on Windows, which manifests when calling
`agrep` or `adist`, the latter being precisely the point of failure in
the FAIL file. See this thread [1] and this post in particular [2] for
more detail. Restoring passing "-Wa,-muse-unaligned-vector-move" to
EOPTS and Makevars solved the problem and check-devel passes. Tomas is
going to be rightfully annoyed with me for forgetting!

Thank you again!!

Avi

[1] https://stat.ethz.ch/pipermail/r-sig-windows/2024q1/000106.html
[2] https://stat.ethz.ch/pipermail/r-sig-windows/2024q1/000113.html

On Sun, Mar 9, 2025 at 1:55 PM Sebastian Meyer <seb.meyer using fau.de> wrote:
>
> Am 09.03.25 um 18:15 schrieb Avraham Adler:
> > Hello.
> >
> > I recently built a new computer and I am now using Windows 11. When
> > building a distribution of R-devel from source using the most recent
> > Rtools44, the build consistently fails at the point it is supposed to
> > build the vignettes from base with the message below. R-4-4-branch
> > (also 87913) completes without error but fails "make check".
> >
> > R-devel Error:
> > ```
> > * DONE (mgcv)
> > building/updating vignettes for package 'grid' ...
> > processing 'displaylist.Rnw'
> > make[2]: *** [Makefile.win:99: vignettes] Error 1
> > make[1]: *** [Makefile:345: vignettes] Error 2
> > make: *** [Makefile:395: distribution] Error 2
> > ```
> > Looking at Makefile.win in .src/library at line 99 shows:
> >
> > ```
> > vignettes:
> > @for pkg in grid parallel utils stats; do \
> >      $(ECHO) "building/updating vignettes for package '$${pkg}' ..."; \
> >      ($(ECHO) "tools:::.install_package_vignettes(\"$(srcdir)/$${pkg}\",
> > \"$(top_builddir)/library/$${pkg}\")") | \
> >          R_DEFAULT_PACKAGES="utils,tools" LC_ALL=C \
> >    R_LIBS_USER=NULL R_LIBS_SITE=NULL \
> >            $(R_EXE) > /dev/null || exit 1; \
>
> The '|| exit 1' only exists since r87822: previously, failures from
> building the base vignettes would be missed.
>
> > done
> > @rm -Rf .vignettes
> > ```
> >
> > While R-4-4 completes "make distribution" compilation, it fails make check with:
> >
> > ```
> > Testing examples for package 'tools'
> >    comparing 'tools-Ex.Rout' to 'tools-Ex.Rout.save' ... NOTE
> >    1062,1063d1061
> >    < Warning in file(con, "r") :
> >    <   file("") only supports open = "w+" and open = "w+b": using the former
> >    1066,1067c1064,1065
> >    <  $ file    : chr ""
> >    <  $ title   : chr ""
> >    ---
> >    >  $ file    : chr "grid.Rnw"
> >    >  $ title   : chr "Introduction to grid"
>
> This output diff from the example in ?tools::vignetteInfo shows that
> your R-4-4 installation indeed lacks base vignettes. Maybe pdflatex
> isn't installed or in the PATH on your new system?
>
> > Testing examples for package 'utils'
> > Error: running examples in 'utils-Ex.R' failed
> > Error: testing 'utils' failed
> > Execution halted
> > make[3]: *** [Makefile.win:29: test-Examples-Base] Error 1
> > make[2]: *** [Makefile.common:208: test-Examples] Error 2
> > make[1]: *** [Makefile.common:193: test-all-basics] Error 1
> > make: *** [Makefile:333: check-devel] Error 2
> > ```
>
> There should be a file utils-Ex.Rout.fail (I think in tests/Examples/).
> What does it say about the error?
>
> Hope this helps!
>
>         Sebastian Meyer
>
> >
> > Some initial research indicated that permission issues can interfere
> > with vignette building. I have set Rtools44 to run as an administrator
> > and I confirmed that my build directory and all its subdirectories are
> > allowed full control by the logged-in windows account.
> >
> > The only major difference between these sets of runs and those over
> > the last decade is the chip (Ryzen 9800X3D), which for vignettes
> > should be irrelevant, and the operating system—Windows 11 Pro instead
> > of Windows 10 Home. I don't know which change from R-4-4 to R-devel
> > triggers the error or why R-4-4 completes and fails the check. I have
> > exhausted my limited knowledge as to how to address the issue. Any
> > ideas as to why this may be happening and what I could do would be
> > greatly appreciated.
> >
> > For completeness:
> >
> > R VERSION AND TOOLS
> > R-devel version: 87913
> > R-4-4-branch:     87913
> > Rtools version: 6459
> >
> > HARDWARE
> > Processor: AMD Ryzen 7 9800X3D 8-Core Processor
> > Installed RAM :192 GB (190 GB usable)
> > System type 64-bit operating system, x64-based processor
> >
> > OPERATING SYSTEM
> > Edition Windows 11 Pro
> > Version 24H2
> > Installed on ‎3/‎7/‎2025
> > OS build 26100.3194
> > Experience Windows Feature Experience Pack 1000.26100.48.0
> >
> > Thank you,
> >
> > Avi
> >
> > ______________________________________________
> > R-devel using r-project.org mailing list
> > https://stat.ethz.ch/mailman/listinfo/r-devel