Issue with remote tables

[robinsones]

The remote table join returns 0 rows, while when collected it returns more than 0.

library(tidyverse)
library(datacampr)
library(funneljoin)

regs <- tbl_views_snowplow_experiment_starts() %>%
  after_inner_join(tbl_views_snowplow_registration_events(),
                   by_user = "domain_userid", 
                   by_time = c("root_tstamp" = "registered_at"),
                   type = "withingap", 
                   dt = as.difftime(30, units = "days")) %>%
  filter(root_tstamp > "2018-07-15")


regs %>%
  collect() %>%
  mutate(user_id = as.integer(user_id)) %>%
  after_join(tbl_main_user_exercises()  %>% 
               filter(completed_at > "2018-08-01") %>%
               mutate(user_id = as.integer(user_id)) %>%
               collect(), 
             by_user = "user_id",
             by_time = c("root_tstamp" = "completed_at"),
             type = "any-firstafter",
             mode = "inner") %>% 
  count()  

regs %>%
  after_join(tbl_main_user_exercises()  %>% 
               filter(completed_at > "2018-08-01"), 
             by_user = "user_id",
             by_time = c("root_tstamp" = "completed_at"),
             type = "any-firstafter",
             mode = "inner") %>% 
  count()  

[robinsones]

These give the same answers though. It seems like it might be an issue with multiple joins in a row

tbl_views_snowplow_registration_events() %>%
  filter(registered_at > "2018-07-15") %>%
  after_join(tbl_main_user_exercises()  %>% 
               filter(completed_at > "2018-08-01"), 
             by_user = "user_id",
             by_time = c("registered_at" = "completed_at"),
             type = "any-firstafter",
             mode = "inner") %>% 
  count()  

tbl_views_snowplow_registration_events() %>%
  filter(registered_at > "2018-07-15") %>%
  collect() %>%
  mutate(user_id = as.integer(user_id)) %>%
  after_join(tbl_main_user_exercises()  %>% 
               filter(completed_at > "2018-08-01") %>%
               collect() %>%
               mutate(user_id = as.integer(user_id)), 
             by_user = "user_id",
             by_time = c("registered_at" = "completed_at"),
             type = "any-firstafter",
             mode = "inner") %>% 
  count()  

[robinsones]

Current fix is not allowing people to do multiple funnel joins in a row remotely ... still need to figure out underlying issue