[Rd] png errors

[Gabriel Becker]

I believe work was done during the r sprint on a Cairo device bug where the
required fix caused unfortunate but expected small changes to output files.
I am not sure if that was merged in but it seems like a pretty big
coincidence if not.

I didnt work on the project directly so others will know more specifics.

Hth
~G

On Sat, Sep 5, 2026, 1:43 PM Terry Therneau via R-devel <
r-devel using r-project.org> wrote:

> I run the test suites of most of the CRAN packages before submitting
> survival with a script that looks a bit like tools:test_packages_in_dir.
> The big difference is that I have a list of things to skip (Bioconductor,
> things that use a GPU, ...) which don't work well on my home box. I tends
> to remove 1 or 2 cycles of submission wrt reverse dependency errors.
>
> In the most recent R-devel versions this approach fails for any package
> that uses PNG, bamlss for example. But if I run R CMD check bamlss*gz from
> the command line all is fine.
> I took the following lines from test_packages_in_dir long ago, which
> stopped things from flashing on my screen, but that appears to no longer be
> enough. Anyone have any hints?
>
> Terry
>
> xvfb_options <- "-screen 0 1280x1024x24"
> pid <- tools:::start_virtual_X11_fb(xvfb_options)
> on.exit(tools:::close_virtual_X11_db(pid), add = TRUE)
>         [[alternative HTML version deleted]]
>
> ______________________________________________
> R-devel using r-project.org mailing list
> https://stat.ethz.ch/mailman/listinfo/r-devel
>

	[[alternative HTML version deleted]]