Closed ajdamico closed 6 years ago
I will definitely consider. Before implementing databases, my hope was that you would be able to just use those functions on srvyr objects with their current syntax. However, databases won't work this way, so it might be worth it.
The stuff from convey would probably come first. I still struggle with how to treat regressions, one of my favorite part of dplyr is that it outputs to a data.frame which made this important to me while creating srvyr. I think broom will be a good guide for regressions, but I don't have a ton of experience with it though.
And would definitely consider a PR :), but the code base is a bit of a mess right now, I hope to do some cleaning so it is either to get through.
Hey @ajdamico, finally got around to allowing extensions (in the rlang branch). I wrote the documentation in vignette form to gather my thoughts, but you're probably the main audience of it, so let me know if you have any questions or suggestions.
hi, do you lay out the recommended usage somewhere, or how would i run the two examples above? thanks
Oops, sorry, thought I put a link last time. Hopefully this helps, but let me know anything doesn't make sense! https://github.com/gergness/srvyr/blob/master/vignettes/extending-srvyr.Rmd
thanks, this is great. probably more practical to convince package authors to add these functions rather than end users.. https://github.com/DjalmaPessoa/convey/issues/267 cc @daob
hi, question, i'm curious if there's any way that survey functions not written by dr. lumley (two examples below) could ever be incorporated into
srvyr
? like, could a user call ofsurvey_gini(...)
look for thesvygini
function? thanks!