Unexpected attribute mapping with missing values in x/y

Dear devs,

it seems like NA values in x/y values (i.e. mapped to coordinates) cause issues (or unexpected behavior) with mapping of other attributes, in this case opacity. I assume this is in-line with #1126.
The attribute seems to be recycled across groups, or only mapped to a single group, depending on how the argument was supplied (referenced as higher and lower level mapping in code).
I was able to reproduce the problem with the code below. Note, that the attribute vector is generated as a column of the data.frame used for plotting.

Cheers!

library(plotly)

n <- 100

# generate test data set with grouping factor z for color
# and index (used for customdata in event_data())
# in shiny
testdat <- data.frame(x = seq_len(n),
                      y = rnorm(n),
                      yy = rnorm(n),
                      z = gl(5, 20),
                      .index = seq_len(n))

# generate attribute vector for opacity
# based on .index (active selection in shiny)
testdat$.opacity <- ifelse(testdat$.index %in% 5:14, 0.9, 0.1)

# assign missing values
testdat$yy[15:19] <- NA

# mapping vector
colvector <- c('1' = "black",
               '2' = "red",
               '3' = "darkorange",
               '4' = "steelblue",
               '5' = "seagreen4")

# "lower level" mapping ---------------------------------------------------

# works as expected
plot_ly(data = testdat) %>% 
    add_markers(type = "scatter",
                x = ~x,
                y = ~y,
                color = ~as.factor(z),
                colors = colvector,
                customdata = ~.index,
                marker = list(opacity = ~.opacity))

# when missing values are present on y-value
# seems to recycle attributes after first group
plot_ly(data = testdat) %>% 
    add_markers(type = "scatter",
                x = ~x,
                y = ~yy,
                color = ~as.factor(z),
                colors = colvector,
                customdata = ~.index,
                marker = list(opacity = ~.opacity))



# "higher level" mapping --------------------------------------------------



# when attribute is mapped at "higher level"
# first "z group" is given opacity attribute,
# but not others
plot_ly(data = testdat) %>% 
    add_markers(type = "scatter",
                x = ~x,
                y = ~y,
                color = ~as.factor(z),
                colors = colvector,
                customdata = ~.index,
                opacity =  ~.opacity)

# when attribute is mapped at "higher level"
# and missing values present,
# opacity not mapped at all
plot_ly(data = testdat) %>% 
    add_markers(type = "scatter",
                x = ~x,
                y = ~yy,
                color = ~as.factor(z),
                colors = colvector,
                customdata = ~.index,
                opacity = ~.opacity)

Session Info:

> sessionInfo()
R version 3.6.3 (2020-02-29)
Platform: x86_64-w64-mingw32/x64 (64-bit)
Running under: Windows 10 x64 (build 18363)

Matrix products: default

locale:
[1] LC_COLLATE=English_United States.1252  LC_CTYPE=English_United States.1252    LC_MONETARY=English_United States.1252
[4] LC_NUMERIC=C                           LC_TIME=English_United States.1252    

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

other attached packages:
[1] plotly_4.9.2       ggplot2_3.3.0.9000

loaded via a namespace (and not attached):
 [1] Rcpp_1.0.4.6      pillar_1.4.4      compiler_3.6.3    later_1.0.0       tools_3.6.3       digest_0.6.25     jsonlite_1.6.1   
 [8] lifecycle_0.2.0   tibble_3.0.1      gtable_0.3.0      viridisLite_0.3.0 pkgconfig_2.0.3   rlang_0.4.6       shiny_1.4.0      
[15] rstudioapi_0.11   crosstalk_1.0.0   yaml_2.2.1        fastmap_1.0.1     withr_2.2.0       dplyr_0.8.5       httr_1.4.1       
[22] vctrs_0.2.4       htmlwidgets_1.5.1 grid_3.6.3        tidyselect_1.0.0  glue_1.4.0        data.table_1.12.8 R6_2.4.1         
[29] farver_2.0.3      purrr_0.3.4       tidyr_1.0.2       magrittr_1.5      scales_1.1.0      promises_1.1.0    ellipsis_0.3.0   
[36] htmltools_0.4.0   assertthat_0.2.1  xtable_1.8-4      mime_0.9          colorspace_1.4-1  httpuv_1.5.2      lazyeval_0.2.2   
[43] munsell_0.5.0     crayon_1.3.4     

Bắt đầu bằng cách chạy bản tái hiện R được cung cấp với plotly.R, so sánh các trường hợp sử dụng y và yy cũng như các ánh xạ opacity ở cấp thấp hơn và cấp cao hơn. Theo dõi cách các giá trị y bị thiếu và các trace được nhóm ảnh hưởng đến các vector thuộc tính, sau đó xác minh rằng opacity vẫn được căn chỉnh với các quan sát và nhóm tương ứng.