r - 下一个行为如何发生(Hmisc 何时/何地导出 `[` )以及在哪里解决它？

Question

该行为涉及 3 个包:Hmisc , psych和 mediation .加载包后 psych和 mediation函数[.describe来自 Hmisc ，但是没有冲突，也没有加载这个包。为什么以及如何发生这种情况？ (实际上我最后可以看到问题出在 loaded via a namespace (and not attached) 行)
让我们首先加载 psych并使用其功能 describe .然后我们对用 [ 生成的对象的一些列进行子集化:

data(sat.act)
ooi <- describe(sat.act)
str(ooi)
Classes ‘psych’, ‘describe’ and 'data.frame':   6 obs. of  13 variables:
 $ vars    : int  1 2 3 4 5 6
 $ n       : num  700 700 700 700 700 687
 $ mean    : num  1.65 3.16 25.59 28.55 612.23 ...
 $ sd      : num  0.478 1.425 9.499 4.824 112.903 ...
 $ median  : num  2 3 22 29 620 620
 $ trimmed : num  1.68 3.31 23.86 28.84 619.45 ...
 $ mad     : num  0 1.48 5.93 4.45 118.61 ...
 $ min     : num  1 0 13 3 200 200
 $ max     : num  2 5 65 36 800 800
 $ range   : num  1 5 52 33 600 600
 $ skew    : num  -0.615 -0.681 1.643 -0.656 -0.644 ...
 $ kurtosis: num  -1.6247 -0.0749 2.4243 0.535 0.3252 ...
 $ se      : num  0.0181 0.0539 0.359 0.1823 4.2673 ...
str(ooi[,c(1,2)])
Classes ‘psych’, ‘describe’ and 'data.frame':   6 obs. of  2 variables:
 $ vars: int  1 2 3 4 5 6
 $ n   : num  700 700 700 700 700 687

直到这里一切顺利。现在让我们加载库 mediation :

library(mediation)
Loading required package: MASS
Loading required package: Matrix
Loading required package: mvtnorm
Loading required package: sandwich
mediation: Causal Mediation Analysis
Version: 4.5.0


Attaching package: ‘mediation’

The following object is masked from ‘package:psych’:

    mediate

search()
 [1] ".GlobalEnv"        "package:mediation" "package:sandwich"  "package:mvtnorm"   "package:Matrix"   
 [6] "package:MASS"      "package:psych"     "tools:rstudio"     "package:stats"     "package:graphics" 
[11] "package:grDevices" "package:utils"     "package:datasets"  "package:methods"   "Autoloads"        
[16] "package:base"    

图书馆 Hmisc尚未加载，但上述相同命令的输出发生了变化:

str(ooi[,c(1,2)])
List of 13
 $ vars    : int [1:6] 1 2 3 4 5 6
 $ n       : num [1:6] 700 700 700 700 700 687
 $ mean    : num [1:6] 1.65 3.16 25.59 28.55 612.23 ...
 $ sd      : num [1:6] 0.478 1.425 9.499 4.824 112.903 ...
 $ median  : num [1:6] 2 3 22 29 620 620
 $ trimmed : num [1:6] 1.68 3.31 23.86 28.84 619.45 ...
 $ mad     : num [1:6] 0 1.48 5.93 4.45 118.61 ...
 $ min     : num [1:6] 1 0 13 3 200 200
 $ max     : num [1:6] 2 5 65 36 800 800
 $ range   : num [1:6] 1 5 52 33 600 600
 $ skew    : num [1:6] -0.615 -0.681 1.643 -0.656 -0.644 ...
 $ kurtosis: num [1:6] -1.6247 -0.0749 2.4243 0.535 0.3252 ...
 $ se      : num [1:6] 0.0181 0.0539 0.359 0.1823 4.2673 ...
 - attr(*, "row.names")= chr [1:6] "gender" "education" "age" "ACT" ...
 - attr(*, "dimensions")= int 13
 - attr(*, "class")= chr "describe"

接下来我尝试调试，我发现它正在使用 [.describe来自 Hmisc 的命名空间:

我试图理解，但我不能:

getNamespaceImports("mediation")$Hmisc
  wtd.var 
"wtd.var" 
is.loaded("[","Hmisc")
[1] FALSE
is.loaded("[.describe","Hmisc")
[1] FALSE
grep("\\[",getNamespaceExports("Hmisc"), value=TRUE)
[1] "[<-.discrete"


sessionInfo()
R version 4.0.3 (2020-10-10)
Platform: x86_64-pc-linux-gnu (64-bit)
Running under: Ubuntu 16.04.7 LTS

Matrix products: default
BLAS:   /usr/lib/atlas-base/atlas/libblas.so.3.0
LAPACK: /usr/lib/atlas-base/atlas/liblapack.so.3.0

locale:
 [1] LC_CTYPE=C.UTF-8       LC_NUMERIC=C           LC_TIME=C.UTF-8        LC_COLLATE=C.UTF-8    
 [5] LC_MONETARY=C.UTF-8    LC_MESSAGES=C.UTF-8    LC_PAPER=C.UTF-8       LC_NAME=C             
 [9] LC_ADDRESS=C           LC_TELEPHONE=C         LC_MEASUREMENT=C.UTF-8 LC_IDENTIFICATION=C   

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

other attached packages:
[1] mediation_4.5.0 sandwich_3.0-0  mvtnorm_1.1-1   Matrix_1.2-18   MASS_7.3-53     psych_2.0.12   

loaded via a namespace (and not attached):
  [1] nlme_3.1-149         lubridate_1.7.9.2    RColorBrewer_1.1-2   numDeriv_2016.8-1.1 
  [5] backports_1.2.0      tools_4.0.3          R6_2.5.0             rpart_4.1-15        
  [9] afex_0.28-0          Hmisc_4.4-2          colorspace_2.0-0     nnet_7.3-14         
 [13] withr_2.3.0          tidyselect_1.1.0     gridExtra_2.3        mnormt_2.0.2        
 [17] emmeans_1.5.2-1      curl_4.3             compiler_4.0.3       cli_2.2.0           
 [21] htmlTable_2.1.0      scales_1.1.1         checkmate_2.0.0      digest_0.6.27       
 [25] stringr_1.4.0        foreign_0.8-80       minqa_1.2.4          rio_0.5.16          
 [29] htmltools_0.5.0      base64enc_0.1-3      jpeg_0.1-8.1         pkgconfig_2.0.3     
 [33] lme4_1.1-26          htmlwidgets_1.5.2    rlang_0.4.9          readxl_1.3.1        
 [37] rstudioapi_0.13      generics_0.1.0       zoo_1.8-8            dplyr_1.0.2         
 [41] ModelMetrics_1.2.2.2 zip_2.1.1            car_3.0-10           magrittr_2.0.1      
 [45] Formula_1.2-4        Rcpp_1.0.5           munsell_0.5.0        fansi_0.4.1         
 [49] abind_1.4-5          lifecycle_0.2.0      stringi_1.5.3        multcomp_1.4-15     
 [53] pROC_1.16.2          carData_3.0-4        plyr_1.8.6           recipes_0.1.15      
 [57] grid_4.0.3           parallel_4.0.3       forcats_0.5.0        crayon_1.3.4        
 [61] lattice_0.20-41      haven_2.3.1          splines_4.0.3        hms_0.5.3           
 [65] tmvnsim_1.0-2        knitr_1.30           pillar_1.4.7         boot_1.3-25         
 [69] estimability_1.3     lpSolve_5.6.15       reshape2_1.4.4       codetools_0.2-16    
 [73] stats4_4.0.3         glue_1.4.2           latticeExtra_0.6-29  data.table_1.13.2   
 [77] vctrs_0.3.5          png_0.1-7            nloptr_1.2.2.2       foreach_1.5.1       
 [81] cellranger_1.1.0     gtable_0.3.0         purrr_0.3.4          assertthat_0.2.1    
 [85] ggplot2_3.3.2        xfun_0.19            gower_0.2.2          openxlsx_4.2.3      
 [89] prodlim_2019.11.13   xtable_1.8-4         class_7.3-17         survival_3.2-7      
 [93] timeDate_3043.102    tibble_3.0.4         lmerTest_3.1-3       iterators_1.0.13    
 [97] cluster_2.1.0        lava_1.6.8.1         statmod_1.4.35       TH.data_1.0-10      
[101] ellipsis_0.3.1       caret_6.0-86         ipred_0.9-9         

谢谢!

Answer 1

这个问题在 R-devel 列表中讨论:https://stat.ethz.ch/pipermail/r-devel/2021-June/080859.html
由 Simon Urbanek 和 Gabriel Becker 解释的答案。
这还不是一个确定的答案，因为它没有讨论完全相同的问题(它不是由第三个包导入的函数，而是一个使包被加载的使用函数)并且它没有详细解释它是如何发生的。