[Rd] Advice on ASAN for gap 1.9

[jinghua zhao]

Dear Ben,

Thank you very much your advice � indeed I was warned of the ragg issue and dropped it out in later versions and somewhat confused with V5 as it does render well on CRAN (sounds like a capacity problem since the other .Rmd has no complaint at all). I am unaware of the docker container and should try that. It is a nice idea to precompute them since the issues are essentially due to vignette building however I need to see if it would increase the package size considerably.

Best regards,


Jing Hua

________________________________
From: Benjamin Schwendinger <benjaminschwe using gmail.com>
Sent: Tuesday, February 17, 2026 22:27
To: r-devel <r-devel using r-project.org>
Cc: jinghua zhao <jinghuazhao using hotmail.com>
Subject: Re: [Rd] Advice on ASAN for gap 1.9

Dear Jing,

you currently have multiple issues.

1st) problem is not directly on your side but lies within ragg as the ASAN error correctly pointed out.

Essentially, ragg/src/./agg/include/agg_font_cache_manager.h:176:21.

On the corresponding line we see

std::memcpy(m_fonts, m_fonts + 1, (m_max_fonts - 1) * sizeof(font_cache*));

Hence, we copy from m_fonts+1 to m_fonts with possible multiple bytes, which is UB. Simplest solution would be to drop `dev = "ragg_png"` from your vignette (as you supposedly also have figured out).

2) You are calling V8 which does not play nicely with the much larger stack frames under ASAN/UBSAN. I would recommend precomputing these outputs.

For reproducing errors regarding ASAN and UBSAN I can fully recommend the rocker containers (rocker/r-devel-san and rocker/r-devel-ubsan-clang) from https://rocker-project.org/images/base/r-devel.html

Best regards,
Ben

Am Di., 17. Feb. 2026 um 20:06 Uhr schrieb jinghua zhao <jinghuazhao using hotmail.com<mailto:jinghuazhao using hotmail.com>>:
Dear R developers,

I am seeking advice regarding some issues of my package from CRAN check. I have made attempts to identify sources of the errors via customized clang/gcc which may not always yield the same errors. I also tried ~/.valgrindrc and --use-valgrind which would say OK but I suspect my ASAN-scheme is fully working. Brefly, the error tracking information can be seen from here,

  *
CRAN checks, https://cran.r-project.org/web/checks/check_results_gap.html
  *
Current attempt. https://cran.dev/gap

For next steps, I wonder if

  *
Anyone has a polished pipelines clang/gcc for Fedora 43  which is what I am using so as to reproduce errors / hints for any additional changes to be made, e.g., https://www.stats.ox.ac.uk/pub/bdr/Rconfig/
  *
To convert the somewhat legacy C/Fortran codes in the package for more modern implementations.

More importantly, any advices will be greatly appreciated.

Best wishes,

Jing Hua

        [[alternative HTML version deleted]]

______________________________________________
R-devel using r-project.org<mailto:R-devel using r-project.org> mailing list
https://stat.ethz.ch/mailman/listinfo/r-devel

	[[alternative HTML version deleted]]