Analyze github repos to identify R packages needing help

[kbroman]

maybe we could identify R packages on github and measure something like number of open issues vs time since last commit. might point to packages of interest that need help

[jdblischak]

@kbroman Good idea! For getting started, @jimhester's package itdepends has some code for gathering relevant info like the number of open issues, etc.

https://github.com/jimhester/itdepends/blob/128cd7e42d866c3beaf8ab40ab3cf2e42392208f/R/github.R#L10

[maurolepore]

👍
Are you thinking about popular, published packages (e.g. tidyverse packages)?
If so, I propose something complementary: To provide a package-help group for chirunconf packages. I'll explain more on a separate issue, but the main idea is to help on the technical side so that folks unfamiliar with package-development tools can realize their ideas and share them as a package at the end of the unconf.

[kbroman]

@maurolepore I wasn't thinking tidyverse particularly, but rather of trying to crawl github to find repositories that were interesting to people (as indicated by there being issues) but not necessarily kept up.

[maurolepore]

Maybe related to #32

[chasemc]

Great idea! I don't think you'll need to crawl, the GH API is pretty good, and: https://github.com/r-lib/gh exists

[jdblischak]

Great idea! I don't think you'll need to crawl, the GH API is pretty good, and: https://github.com/r-lib/gh exists

The Search API is more restrictive than the other parts of the API. I often get rejected "for triggering abuse mechanisms" even though I am only querying a few hundred results. Searching every R package on GitHub would take some time.

Another possibility would be to start with this curated list of GitHub R packages. Starting from this list, then we could use the GitHub API to query specific attributes about each repository.

[jimhester]

The GH API lets you search repositories by number of help-wanted-issues, which might be a way to go to find some places to help.

search <- gh::gh("/search/repositories", q = "language:r", sort = "help-wanted-issues", order = "desc")

library(purrr)
library(tibble)

map_chr(search[[3]], "full_name")
#>  [1] "UptakeOpenSource/uptasticsearch"              
#>  [2] "BioinformaticsFMRP/TCGAbiolinks"              
#>  [3] "cbeleites/hyperSpec"                          
#>  [4] "Huh/PopR_SDGFP"                               
#>  [5] "International-Soil-Radiocarbon-Database/ISRaD"
#>  [6] "TommyJones/textmineR"                         
#>  [7] "UptakeOpenSource/pkgnet"                      
#>  [8] "ropenscilabs/learngganimate"                  
#>  [9] "slowkow/ggrepel"                              
#> [10] "BonnyCI/ci-plunder"                           
#> [11] "OpenMx/OpenMx"                                
#> [12] "ProvTools/provR"                              
#> [13] "statnet/ergm"                                 
#> [14] "retrography/OrientR"                          
#> [15] "trestletech/plumber"                          
#> [16] "ices-tools-prod/fisheryO"                     
#> [17] "jackwasey/icd"                                
#> [18] "rich-iannone/DiagrammeR"                      
#> [19] "ropensci/tabulizer"                           
#> [20] "cloudyr/googleComputeEngineR"                 
#> [21] "TIBHannover/BacDiveR"                         
#> [22] "HenrikBengtsson/aroma.seq"                    
#> [23] "HenrikBengtsson/Wishlist-for-R"               
#> [24] "theclue/facebook.S4"                          
#> [25] "NKU-DSC/RTrainingMaterials"                   
#> [26] "kevinwolz/hisafer"                            
#> [27] "fabian-s/tidyfun"                             
#> [28] "vertica/DistributedR"                         
#> [29] "ropenscilabs/dataspice"                       
#> [30] "franzbischoff/tsmp"

^{Created on 2019-03-08 by the reprex package (v0.2.1)}

Of course this depends on the repo owners using that specific tag for issues, which many do not.