Unicode text fails after attaching ragg

[StefanoMezzini]

Attaching the ragg package prevents one from using unicode characters in base plot() and ggplot(), as any string with unicode will be blank. However, printing a plot (with either plot() or ggplot()) before attaching ragg fixes the issue:

library('ragg')
plot(1, xlab = '\U1D53C doesn\'t work')
ggplot2::ggplot() +
  ggplot2::geom_point(ggplot2::aes(1, 1)) +
  ggplot2::xlab('\U1D53C doesn\'t work')

# restart R (Ctrl + Shift + F10)

plot(1, xlab = '\U1D53C')
library('ragg')
plot(1, xlab = '\U1D53C now works')
ggplot2::ggplot() +
  ggplot2::geom_point(ggplot2::aes(1, 1)) +
  ggplot2::xlab('\U1D53C now works')

# restart R (Ctrl + Shift + F10)

ggplot2::ggplot() +
  ggplot2::geom_point(ggplot2::aes(1, 1)) +
  ggplot2::xlab('\U1D53C')
library('ragg')
plot(1, xlab = '\U1D53C now works')
ggplot2::ggplot() +
  ggplot2::geom_point(ggplot2::aes(1, 1)) +
  ggplot2::xlab('\U1D53C now works')

I am using R 4.2.1, RStudio 2022.07.1+554 "Spotted Wakerobin", and ragg 1.2.3 on Windows 10.

[thomasp85]

I'm not able to reproduce this issue...

What do you get when you run systemfonts::system_fonts()

[StefanoMezzini]

Thanks for the quick response! I've attached a csv with the output from systemfonts::system_fonts(). The issue occurs on both my laptop and my work computer.
fonts.csv

[thomasp85]

Hi @StefanoMezzini — are you still experiencing this. If so, can I get you to provide a session info as well as let me know where you are observing the behaviour. The code snippet you provided indicates that you see it in the plot pane in RStudio / the graphics window in the R gui

[StefanoMezzini]

Hi @thomasp85, I still have the same issue. Here is my session info before attaching ragg:

R version 4.2.2 (2022-10-31 ucrt)
Platform: x86_64-w64-mingw32/x64 (64-bit)
Running under: Windows 10 x64 (build 19045)

Matrix products: default

locale:
[1] LC_COLLATE=English_Canada.utf8  LC_CTYPE=English_Canada.utf8    LC_MONETARY=English_Canada.utf8 LC_NUMERIC=C                   
[5] LC_TIME=English_Canada.utf8    

attached base packages:
[1] stats     graphics  grDevices utils     datasets  methods   base     

loaded via a namespace (and not attached):
[1] compiler_4.2.2 tools_4.2.2

and after attaching ragg:

R version 4.2.2 (2022-10-31 ucrt)
Platform: x86_64-w64-mingw32/x64 (64-bit)
Running under: Windows 10 x64 (build 19045)

Matrix products: default

locale:
[1] LC_COLLATE=English_Canada.utf8  LC_CTYPE=English_Canada.utf8    LC_MONETARY=English_Canada.utf8 LC_NUMERIC=C                   
[5] LC_TIME=English_Canada.utf8    

attached base packages:
[1] stats     graphics  grDevices utils     datasets  methods   base     

other attached packages:
[1] ragg_1.2.4

loaded via a namespace (and not attached):
[1] compiler_4.2.2    tools_4.2.2       rstudioapi_0.14   textshaping_0.3.6 systemfonts_1.0.4

If I run plot(1, xlab = '\U1D53C') first, then the session info becomes:

R version 4.2.2 (2022-10-31 ucrt)
Platform: x86_64-w64-mingw32/x64 (64-bit)
Running under: Windows 10 x64 (build 19045)

Matrix products: default

locale:
[1] LC_COLLATE=English_Canada.utf8  LC_CTYPE=English_Canada.utf8    LC_MONETARY=English_Canada.utf8 LC_NUMERIC=C                   
[5] LC_TIME=English_Canada.utf8    

attached base packages:
[1] stats     graphics  grDevices utils     datasets  methods   base     

loaded via a namespace (and not attached):
[1] compiler_4.2.2  tools_4.2.2     rstudioapi_0.14

and attaching ragg now makes it

R version 4.2.2 (2022-10-31 ucrt)
Platform: x86_64-w64-mingw32/x64 (64-bit)
Running under: Windows 10 x64 (build 19045)

Matrix products: default

locale:
[1] LC_COLLATE=English_Canada.utf8  LC_CTYPE=English_Canada.utf8    LC_MONETARY=English_Canada.utf8 LC_NUMERIC=C                   
[5] LC_TIME=English_Canada.utf8    

attached base packages:
[1] stats     graphics  grDevices utils     datasets  methods   base     

other attached packages:
[1] ragg_1.2.4

loaded via a namespace (and not attached):
 [1] fansi_1.0.3       assertthat_0.2.1  dplyr_1.0.10      utf8_1.2.2        grid_4.2.2        R6_2.5.1          DBI_1.1.3         lifecycle_1.0.3  
 [9] gtable_0.3.1      magrittr_2.0.3    scales_1.2.1      ggplot2_3.4.0     pillar_1.8.1      rlang_1.0.6       cli_3.4.1         rstudioapi_0.14  
[17] generics_0.1.3    vctrs_0.5.0       textshaping_0.3.6 tools_4.2.2       glue_1.6.2        munsell_0.5.0     compiler_4.2.2    systemfonts_1.0.4
[25] pkgconfig_2.0.3   colorspace_2.0-3  tidyselect_1.2.0  tibble_3.1.8 

I've attached the plots that result from the first two calls of plot() and ggplot() from the code in my first comment. I am viewing the plots in the RStudio plot pane, but I get the same issue if I call png() or pdf() after attaching ragg without running plot(1, xlab = '\U1D53C') first. However, when I try saving the image to pdf, I get the warnings

Warning messages:
1: In title(...) :
  conversion failure on '𝔼 doesn't work' in 'mbcsToSbcs': dot substituted for <f0>
2: In title(...) :
  conversion failure on '𝔼 doesn't work' in 'mbcsToSbcs': dot substituted for <9d>
3: In title(...) :
  conversion failure on '𝔼 doesn't work' in 'mbcsToSbcs': dot substituted for <94>
4: In title(...) :
  conversion failure on '𝔼 doesn't work' in 'mbcsToSbcs': dot substituted for <bc>
5: In title(...) :
  conversion failure on '𝔼 doesn't work' in 'mbcsToSbcs': dot substituted for <f0>
6: In title(...) :
  conversion failure on '𝔼 doesn't work' in 'mbcsToSbcs': dot substituted for <9d>
7: In title(...) :
  conversion failure on '𝔼 doesn't work' in 'mbcsToSbcs': dot substituted for <94>
8: In title(...) :
  conversion failure on '𝔼 doesn't work' in 'mbcsToSbcs': dot substituted for <bc>

and the pdf I attached.

[thomasp85]

So, you are experiencing issues, not with the ragg devices but with the default devices, and those issues seem to appear after you attach ragg?

[StefanoMezzini]

I believe so? I am only using the default figure pane with the code in the first message, and I have updated ragg to 1.2.5.9000.

[thomasp85]

And if you go into "global options">"general">"graphics", what does it say as the backend?

[StefanoMezzini]

I have it set to (Default). Windows also fails, and Cairo and Cairo PNG produce boxes but no special characters, while AGG succeeds.

[thomasp85]

@pmur002 this issue along with #35 is completely baffling to me

Somehow, loading textshaping interferes with the text rendering in various ways. Do you have any insight into what might cause this?

[jeroen]

Can you try installing the latest dev versions and test if the problem still exists?

install.packages("textshaping", repos = "https://r-lib.r-universe.dev")

[pmur002]

Off the top of my head, the only thing this reminds me of is FontConfig caching producing the wrong glyphs (in a very complicated set up).

[thomasp85]

Thanks - this is unfortunately happening on windows so FontConfig doesn't play in