Embrace the tibble! #250
Embrace the tibble! #250
Conversation
Did not make changes to internals: 1. They rely on rownames. 2. They are complicated.
|
Main benefits:
|
Codecov Report
@@ Coverage Diff @@
## master #250 +/- ##
=====================================
Coverage 100% 100%
=====================================
Files 64 64
Lines 4468 4486 +18
=====================================
+ Hits 4468 4486 +18
Continue to review full report at Codecov.
|
|
R/commands.R:91:8: style: Place a space before left parenthesis, except in a function call. for(attribute in c("srcref", "srcfile", "wholeSrcref")){
^ |
|
Maybe use hms for durations? |
|
I just tried. It might work if I did it differently, but it seems to turn the times into NA's. load_basic_example()
make(my_plan)
bt <- build_times()
library(lubridate)
bt$elapsed <- hms(bt$elapsed)
## In .parse_hms(..., order = "HMS", quiet = quiet) :
## Some strings failed to parse, or all strings are NAs
head(bt)
## item type elapsed user system
## 1 coef_regression1_large target <NA> 0.004s 0s
## 2 coef_regression1_small target <NA> 0.004s 0s
## 3 coef_regression2_large target <NA> 0.008s 0s
## 4 coef_regression2_small target <NA> 0.004s 0s
## 5 data.frame import <NA> 0.008s 0s
## 6 knit import <NA> 0.06s 0s |
|
This works for me: bt$elapsed %>% lubridate::as.difftime() %>% hms::as.hms()I wonder if we can change hms so that the first step in the pipe is not required. At this point it seems better to add pillar support to lubridate (or the other way round). |
|
It works for me too, but I'm not sure I like the output format: > bt$elapsed %>% lubridate::as.difftime() %>% hms::as.hms()
## 00:00:00.008
## 00:00:00.004
## 00:00:00.004
## 00:00:00.007
## 00:00:00.008
## 00:00:00.010 |
|
The dev version of pillar now uses |
Now used for workflow plan data frames and the output of
dataframes_graph(). Supports #169 and #247. Remaining challenges:tibble, so some internals around staged parallelism will need careful refactoring first.