Commit
This commit does not belong to any branch on this repository, and may belong to a fork outside of the repository.
R-dplyr: Autospec creation for update from version 0.8.5 to version 1…
….0.0
Adam Davis (1):
Proofread and update documentation for *_join functions (#4816)
Alexandre Courtiol (1):
Prevent collapse of data.frame to vector in across (#4867) (#4868)
Anthony Damico (1):
Fix oversight in example text (#4832)
Ben Schneider (1):
Inside `dplyr:::tbl_if_vars()`, use `pull()` instead of base column indexing (#4879)
Bill Denney (1):
Allow group_modify() to work with ... args (Fix #4509) (#4620)
Binod Jung Bogati (1):
Removed rbind_list and rbind_all (#4474)
Christian Heinrich (1):
Update a comment in window-functions.Rmd (#5238)
Christophe Dervieux (1):
typo - correct package name (#4838)
Colin Rundel (1):
Improve error message formatting in bad_eq_ops() (#4476)
Dan Chaltiel (1):
Check return type of predicate in tbl_if_vars (#5003)
Danielle Navarro (1):
Updates documentation to filter() (#4820)
Danielle Quinn (1):
Refresh `select()` docs (#4823)
Danielle Smalls (1):
update computational backend links (#4817)
Davis Vaughan (24):
More efficient replacement of `NULL` with `NA_integer` in `join_rows()` (#4880)
Avoid `c()` calls when only `x` rows are required (#4881)
Use `new_data_frame()` to avoid materializing row names (#4882)
Generalize avoidance of `c()` calls to right and full joins (#4884)
Use `names2()` to guard against `NULL` column names (#4896)
Fix sequential updating (#4894)
Pass on the number of rows to `new_tibble()` (#4900)
Use `new_tibble()` in `$pick()` for faster construction (#4901)
Swap `eval_tidy()` for `env_get_list()` (#4908)
Cache `eval_select()` results in `across()` on a per expression basis (#4909)
Update to latest actions workflow (#4944)
Update test files to lifecycle 0.2.0 warning messages (#4951)
Use `unlist()` in `join_rows()` until `vec_c()` performance improves (#4975)
Avoid `vec_assign()` call when all rows match (#4976)
Use a double loop to apply functions in `across()` (#4978)
Use `vec_is_list()` in `bind_cols()` and `bind_rows()` (#4990)
Maintain grouping order in grouped_df methods (#5045)
Don't `.` prefix `cols` in `c_across()` (#5021)
Simplify `dplyr_reconstruct()` (#5160)
Refactor `group_split()` to use `dplyr_row_slice()` (#5167)
Simplify `group_split()` using `ungroup()`, not `as_tibble()` (#5175)
Ensure that `arrange()` always calls `dplyr_row_slice()` (#5205)
Ensure that `filter()` always calls `dplyr_row_slice()` (#5206)
Correctly register rowwise reconstruct method (#5229)
DavisVaughan (13):
Add `current_vars()` to lookup the current column names
Pull names from updated bindings in `cur_data()`
Use the names of `$resolved` to ensure the order is correct
Add `current_cols()` for low overhead extraction
Tweak `across()` to work sequentially without resolving unused columns
Push `across()` column selection into the mask
Simplify `cur_data()` with `$current_non_group_vars()`
Improve comment about how `across_cols()` is used
Remove `struct vctrs_arg*` argument from `short_vec_recycle()`
Bump minimum vctrs version
Add regression test for `bind_rows()` and named factors
Add a Remote on the relevant rlang PR
Switch to just dev rlang
Enrico Spinielli (1):
typo (double 'because') (#4957)
Eric Stern (1):
Refresh `bind()` docs (#4833)
Greg Freedman Ellis (1):
use n as default wt in tally (#4581)
Gustav W Delius (1):
Fix documentation typo (#4647)
Hadley Wickham (216):
Add section on backends to readme (#4490)
Make `group_map()` a generic (#4583)
Implement tbl_ptype (#4584)
Re-document
Eliminate unnused inheritParam tags
Use generics
Re-export set operations from generics
Update to roxygen2 7.0.1 (#4630)
Ensure all functions listed on reference page
Handle missing entries in starwars data
Remove skipped test
use github actions
Simplify failing test
Skip two encoding issues on windows
Don't require all suggested packages to be installed
Use travis only for pkgdown
Eliminate appveyor
Update badges in README
Fix yaml parse error
Re-document
Renable vignette building
Use lifecycle package and deprecate funs()
Remove tbl_cube() and nasa (#4658)
Remove all.equal.tbl_df method (#4657)
Move all lazy eval functions into a single file (#4659)
Split funs() and as_funs_list() into own files
Eliminate/rewrite tests using funs
Remove unnused astyle test
Selectively ungroup variables (#4671)
Remove duplicated news entries
Give hint in order_by() if used like arrange()
count()/tally() now errors if default name already present in output
First pass at methods_rd()
Rewrite help_topic to use low-level approach
More iteration
Start working on wording for doc with single verb
Label argument that use tidy evaluation, tidy select, and dynamic dots (#4684)
Deprecate bench, compare, and eval functions (#4696)
Preserve order of variables in transmute (#4694)
Improvements to distinct (#4691)
Move methods next to generics (#4690)
Improve summarise docs
Improve arrange docs
across() tweaks
New family of slice helpers (#4687)
Drop lengths backport
Delete out of date API reference
Make failwith defunct
failwith is defunct, not deprecated
id() is defunct
Document id() and failwith() together
Standardise deprecations (#4706)
Tweaks to distinct docs
Match test-joins.R to join.R
Deprecate group_by(..., add = TRUE) (#4701)
Deprecate src_local() (#4707)
Tibble deprecations (#4705)
Add some missing grouped_df methods (#4710)
Allow modification of grouping_vars (#4712)
Add names<-.grouped_df method
Upgrade tidyselect (#4720)
Verb genericity (#4715)
Trim and reorganise arrange() tests
Add return type tests
A little tidying of mutate tests
Re-arrange grouped-df.R contents
Clarify grouped_df()
Re-document
Remove unnused match_vars() function
Fix accidental regression
group_by(.dots) is deprecated
Pin tibble to known good version
Simplifying build_groups a little (#4727)
Tweak error message
Simplify group metadata (#4728)
Rename context functions to common prefix (#4748)
Default cols = everything()
Rewrite all join functions (#4741)
Consistent names for verb helpers
Remove code doubled via accidental merge
Improve documentation of output type
Minimise regrouping work (#4751)
Refine package docs
Separate select() and rename() functions
Mark all_equal() as questioning
Eliminate expect_groups() helper
Tease apart documentation for join functions
Fix thinko in generics doc
Revise old tidy eval docs
Inherit params from arrange()
Note about RStudio's behaviour
Add extra line break
Don't link to self
Apply pattern to all major verbs
Polish titles and of primary verbs
Document join methods with generics
Simplify pkgconfig usage
Improve section names
Document slice helpers with slice
News tweaking
Consistently place badge at top of description
WS
Explain experimental/questioning/superseded
with_groups is experimental
Use lowercase rd for pkgdown compatibility
Tweak reference index for condense/rowwise
mutate() doc tweaks
Emphasise that rowwise(...) is different to group_by(...)
Improve mutate()/summarise() multi-row/multi-col docs
Mark scoped helpers as superseded (#4769)
Move DataMask to data-mask.R
Redocument
Implement relocate
Work around buglet in rlang
New programming with dplyr vignette (#4724)
Improvements to contextual functions (#4772)
Simplify filter() C implementation
Simplify filter C code
Remove out dated article
Rename dplyr_tidy_eval to dplyr_data_masking
Simplify glue_c()
Skip failing test for now
Redocument
Give more control over which variables mutate() should keep (#4773)
Remove summarise tests
Roughly group summarise tests
More re-org and tweaking
Continue to refactor summarise() tests
Test 0-row group case
Control position of new mutate() columns (#4774)
Divide news into groups
Polishing news
Re-document
Weaken list_of() tests so they pass with dev vctrs
Carefully simplify bind implementations
Correct binds test file name
Restore list-binding behaviour
Unpack tibbles in bind_cols()
Switch to starwars dataset in intro
Default slice helpers to use n = 1
Add missing words
WS
Flag recode() as questioning
Remove implicit NA warning
Polish output descriptions
Prevent relocate from losing non-contiguous variables
Add missing news bullet
Remove coercion rules section from two-table vignette
Minor relocate() simplification
Programming vignette tweaks
Review count docs
Move lobstr to suggests
Prune unused suggested packages
Remove bit64 and hms dependencies
use_tidy_description()
Remove duplicate rlang remote
Can use CRAN pillar
group_cols() is broken with new tidyselect approach (#4730)
Coerce both x and y keys to same type
Fix misunderstanding of tidyselect
Be more precise about cur_column()
Row-wise and col-wise vignettes (#4902)
Switch to github actions for website
Add section on breaking changes (#4932)
Fixing failing tests on R-devel
Implement c_across() (#4924)
Advertise breaking change about n() and row_number() (#4918)
Deprecate progress_estimated()
Fix nest_join() bug
Implement rename_with() (#4923)
Use CRAN tibble (#4937)
Remove out dated package docs
Only install dev bench for benchmarks
R CMD install is not an R command :|
Add missing arg
Implement na_matches = "never" (#4940)
Flag .before and .after as experimental in the docs
Supersede rather than deprecate do()
Organise lifecycle section of news
Better news for vignette updates
Deprecate src_sqlite() and friends (#4941)
Use latest lifecycle conventions (#4946)
Fix deprecation warning
Fix mistakes in programming vignette
Use length.out instead of length
Add superseded badge
Tweak news for vctrs coercion rules
Fix merge issue
Add news bullet
Updated gitignore
Re-export all_of() and any_of()
Tweak gotcha wording
Improve hint for old grouping (#5052)
Nesting interface changes (#4926)
Move check for class before coercion (#5053)
Revert "Don't count regroups in tests (#5061)" (#5076)
Coerce join inputs into tibbles (#5059)
Simplify add_computed_columns() implementation
Improve website navbar + article index (#5081)
Fix computation of proportion in slice_max()
Improvements to dplyr extensibility (#5169)
Adjust tests for waldo (#5121)
Get count() working in database backends (#5114)
Re-run revdeps
Tweak c_across() docs
Add ... to pull() (#5226)
More documentation hints about rowwise_df
Close database connection in test
Convert coalesece to use vctrs (#5187)
Use where() in more tests
nest_by() improvements (#5171)
Update dplyr logo (#5248)
Re-run revdeps
More improvements to count() (#5276)
Update CRAN comments
Increment version number
Hiroaki Yutani (2):
Fix a typo (#4785)
Fix function names in lifecycle::deprecate_warn() (#4948)
Ian Cook (3):
Add dots to generic src_tbls() arguments (#4485)
Improve NEWS bullet for #4485 (#4672)
Implement keep = T in left_join(), right_join() (#4762)
Ilari Scheinin (2):
pull() can also return named vectors (#4102)
Update NEWS.md for PR #4102
Jane Wall (1):
re-organized and deleted some sections. (#4821)
Jen Ren (1):
Proofread programming vignette (#4819)
Jennifer (Jenny) Bryan (2):
Fix the r-devel job (#5237)
Refactor encoding test helpers and use withr::local_locale() (#5236)
Jenny Bryan (1):
Make cache keys match (and bump, for good measure)
Jim Hester (1):
Set RSPM separately from CRAN
JohannesNE (1):
Fix 3 typos in rowwise vignette (#4956)
Jon Harmon (1):
Expanded join help for multiple variables (#4870)
Kirill Müller (16):
Fix tests
Use pluck() instead of map()
Review documentation for naming in scoped functions (#4452)
Use as_label() for determining name of RHS table after nest_join() (#4651)
WIP: Fix character encoding problems (#4453)
Fix broken test (#4743)
Set tibble version (#4745)
Tweak tests (#4920)
Fix protection error in corner case (#4945)
Update roxygen to 7.1.0
Update roxygen to 7.1.0
Un-buildignore files named bench (#4985)
ntile() uses database logic: larger bins come first (#5054)
Soft-deprecation message for using _if()/_at()/_all() with a quosure (#4933)
Avoid unname() on tibbles (#5159)
rows_() (#5142)
Laura Wiley (1):
Update lead + lag documentation (#4818)
Lionel Henry (22):
Fix defusal of tidyselect arguments
Depend on CRAN versions
Remove redundant length check
Clean up NEWS
Fix internal names in `bind_rows()` test
Don't count regroups in tests (#5061)
Depend on CRAN rlang
Slightly clearer formulation in `?select`
Inherit tidyselect documentation
Include overview in description
Include overview in <tidy-select> topic
Update documentation and tests to use `where()`
Use glue syntax for programming with names
Depend on CRAN tidyselect
Work around doc import issues
Fix pkgdown build
Use `where()` in remaining documentation (#5259)
Return data frame from `vec_data()` (#5271)
Fix `arrange()` issue with unruly class
Test that unruly class can be filtered
Strip attributes before reconstruction (#5277)
Remove error classes for now (#5280)
Mara Averick (2):
Use https for all URLs in DESCRIPTION
Update License year
Maria Paula Caldas (1):
Add example in recode about NA types (#4472)
Mark Dulhunty (1):
rowwise vignette last line typos (#4958)
Matthew J. Michalska-Smith (1):
Update group_split.R (#5143)
Megan Beckett (1):
Improve Starwars data set gender and sex classifications of characters (#4480)
Mike K Smith (1):
Create group by vignette (#4829)
Mikko Marttila (1):
Prefix across() args with dots and pass ellipsis after .fns (#4979)
Mine Cetinkaya-Rundel (3):
Vignette rename: row-wise and col-wise (#5104)
Minor formatting edits on dplyr vignette (#5102)
Minor edits (#5107)
Mine Çetinkaya-Rundel (8):
Simplify data
Remove rounding with across()
Use pipe to clarify usage of rowwise
Grammatical
Taking sqrt so it should be rmse
Swap list columns and repeated function calls sections
Remove redundant word
Plural
Pedro Perez (1):
Deprecate select_vars, rename_vars, select_var, current_vars (#4470)
Rebecca Janis (1):
Refresh distinct docs (#4815)
Romain Francois (117):
Increment version number
vars_select_syms() converts dplyr_sel_vars to character before callling the function(s)
comment out the travis job against devel packages
group_walk() returns its result invisibly.
`group_by()` does not create an arbitrary NA group when grouping by factors with `drop = TRUE`.
select_if() discards column when predicate evaluates to NA.
Better performance for slicing factors and ordered factors.
url update
revdep results 0.8.4 [ci skip]
Remove replaceable uses of R_UnboundValue since it's not part of the official R api, use a sentinel instead
cache base::rm
calling back to R to evaluate base::rm()
update revdep results
update cran-comments
adding 0.8.4 with future blog post in pkgdown News section
using Rcpp::List instead of ListView, seems to fix the revdep probmem with bayesCT
explicit copy constructor for GroupedSlicingIndex
not using unsafe Rcpp::Environment::namespace_env().
merge hickup
Consistently use list_of(integer()) for the rows
unskip and update test for group_by() and POSIXlt
Simplify Makevars[.win]
Use obj_addrs? from lobstr:: instead of reimplementing them here.
unskip test
unskip test
skipping NA in automatic by matches in joins
dev tibble does not allow matrix like indexing
only remove variable if present in the first place
using dev tibble
skip test
adapt to latest changes in vctrs, esp vec_group_pos()
brind back regression test for #1578 because vctrs can cope, i.e. r-lib/vctrs#540
skip test about summarise() keeping names in result until conversation is resolved in tidyverse/dplyr#2675
update joins test to changes in tibble
unskip test after tidyverse/tibble#626 was merged
unskip after r-lib/vctrs#540
ok for summarise() to return named columns. closes #2675
unskip test which now passes
unskip test
unskip after tidyverse/tibble#627
remove unneeded test bc undesirable automatic coercion from raw to integer
unskip #636 related test
unskip #1402 related test
handling the case when some results are NULL in a mutate().
unskip #4186 test
fix bind_rows() yet again
update tests
reskip test
rebase hickup
transmute() no longer warns when a variable has been set to NULL
bind_cols() handles packs
switch order of remotes
using r-lib/vctrs#688
deleting R/RcppExports.R which should no longer be here already
Adding regression test for #2326
regression test from #4516
Adding regression test for #2457
tests for #3630
Prefer using R_xlen_t when possible.
switch back to master branch of vctrs
+ bench script
filter() handles named logical vectors as results.
Not testing for size too early in filter()
internalize calculation of chunks in mutate()
internalize filter() sprinkling
added internal dplyr_mask_eval_all() used in slice()
global variables used by %<-%
pointer not initialized
No longer using assertthat
Separate production of new columns from reconstruction of the data in mutate()
rework transmute() so that it uses the code extracted out of mutate()
No longer need the .select_grouped_df() internal function
+test for transmute() handling of auto splicing
s/Rf_error/Rf_errorcall(NULL/
DataMask$add() needs to resolve the chunks promise. closes #4677
document() after #4833 merge
library(dplyr, warn.conflicts = FALSE) on bench scripts.
script to run bench scripts on 0.8.3 and current branch.
revdep check results
revdep update
first pass at sorting revdep findings
rlang 0.4.5 is on CRAN
revdepcheck results 2020-03-10
revdep update 2020-03-11
vctrs 0.2.4 is on cran 🎉
some revdep updates
revdep update
revdep update after #5004
test was empty
revdep update
⏪ to a working version of vctrs.
use master version of vctrs again
revdep update
revdep update
Mention the `Input must be a vector, not a ...` in the Breaking changes
rev dep update
unskip test, as r-lib/vctrs#776 is dealt with
remove unused function
revdep update
revdep update
revdep update
revdep update
revdep update
revdep update
revdep update
support R 3.2 again
-Wsign-compare
vctrs 0.3.0 is on CRAN 🎉
typo.
devtools::build_readme()
skip test for now
revdep update
minor NEWS edit to mention where()
revdep update
revdep_report_cran() + edits
revdep update
update cran-comments
Romain François (99):
vctrs powered group_by() (#4504)
vctrs powered distinct() and n_distinct() (#4512)
vctrs based set operations (#4515)
vctrs:: based group_split() (#4514)
vctrs:: based arrange() (#4521)
vctrs:: based summarise() filter() slice() mutate() (#4523)
vctrs:: based joins (#4546)
vctrs:: based versions of bind_*() (#4562)
another special case to fix bind_rows() so that purrr::map_df(mtcars, mean) does what is expected (#4591)
Eliminate dependency on Rcpp (#4619)
all.equal() works on nameless tibbles. (#4573)
group_split() returns a vctrs_list_of (#4623)
vctrs:: based implementation for lead() and lag(), e.g. respcting bit64::integer64 classes (#4633)
internal implementation for validate_grouped_df (#4640)
faster internal group_keys() (#4642)
Performance improvements for summarise() (#4646)
Eliminate unneeded (and broken) method override (#4665)
row order for left_join() and right_join() (#4669)
across() (#4586)
Fix encoding problems with joins and data mask
group_by() and distinct() understand auto splicing (#4676)
`filter()` handles data frame results when all columns are logical ve… (#4681)
filter() allow named input when not calls (#4666)
unskip tests (#4674)
Less use of STL (#4698)
across() selection does not include grouping variables (#4700)
avoiding a NOTE with %<-% (#4714)
rename before vec_rbind() for the by_x part of the full join (#4713)
Using verify_output() instead of expect_error() (#4703)
stop_dplyr() experiment (#4726)
mutate() can "remove" a variable that did not exist (#4732)
summarise() tidy recycling (#4733)
Bullet errors for summarise() (#4729)
bullet errors for mutate() (#4747)
reduce redundant code with a few macros (#4756)
Fix cur_group() (#4758)
Revised rowwise() + condense() (#4760)
Detailed information about the keys of a group on error (#4746)
row wise specific error bullets (#4783)
test arrange() support for across() (#4786)
Add names= argument to across() (#4782)
For ungrouped data frames, chunks are always the whole column, we can short circuit vec_slice() but we have to keep the active binding for track_usage (#4809)
Remove warning about unused n_i (#4835)
Only one defineVar for current_group (#4836)
Using promises instead of active bindings (#4812)
Additional rowwise() magic within the same mutate() call. (#4846)
Faster internal group_keys implementation (#4854)
join_mutate() using vec_ptype_common() instead of vec_ptype2() (#4863)
tweaks to arrange() errors (#4807)
mutate() performance improvements (#4858)
setup continuous benchmarks (#4862)
across() passes ... to functions (#4890)
Remove dependency on zeallot (#4891)
adapt to r-lib/vctrs#855 (#4906)
arrange() with 0 quosures returns input (#4961)
glue:: is on its way to CRAN, and that fixes a few problems identified in revdeps (#4969)
mutate() skips chop/unchop with symbol expressions (#4973)
recalculate which_used on removes. (#4981)
glue 1.3.2 is on CRAN, no longer need to be on remotes
cache more in across_setup() (#4982)
bind_cols(<...>) -> tibble (#5002)
using dev vctrs so that dplyr can depend on r-lib/vctrs#931 (#5004)
protect against tibble() .data masking in older versions of tibble:: (#5009)
grouped_df(drop=) defaults to use group_by_drop_default(data) (#5007)
use [, 0] instead of [0] (#5024)
Support for = <symbol> in mutate() (#5031)
vars() no longer using lifecycle:: (#5013)
join_rows() giving meaningful error message (#5025)
group_map() not warning about keep= for rowwise data frame. (#5049)
use vec_sze() instead of length() in lead() and lag() (#5050)
Remove extra space before ":" from error messages (#5047)
use vec_rep() instead of deprecated vec_repeat() (#5056)
grouo_by() with mutate() semantics (#5057)
N distinct eval tidy (#5072)
specific error for (lag|lead)(default=) when incompatible type. (#5075)
DataMask$resolve() (#5080)
ignore data frames with 0 columns as results in summarise() (#5088)
bind_rows(.id = NULL) does not set names. (#5089)
Changes in some messages from vctrs (#5093)
rowwise() as a generic. (#5152)
error messages improvements (#5099)
lead(), lag(): cast default to common type rather than to type of x. (#5181)
Group data init (#5193)
across() and c_across() when used out of dplyr context (#5194)
summarise(.groups=) (#5145)
Update gitHub actions (#5188)
Various errors fixes (#5197)
dplyr_col_modify.data.frame() calling dplyr_reconstruct() instead of dplyr_reconstruct.data.frame() (#5198)
vec_c(.name_spec = zap()) from r-lib/vctrs#1091 (#5201)
group_split(keep=) -> group_split(.keep=) (#5191)
tally() not using .groups= (#5203)
stop_dplyr() gains a parent= that can be used when rethrowing e.g. a vctrs error (#5219)
across(<empty>) -> data frame with 0 columns and 1 row. (#5232)
round down proportions on slice_tail() (#5234)
summarise() regrouping information, take 2 (#5249)
extra care when calling `[` for a data frame (#5245)
swapi.co -> swapi.dev (#5270)
Findings from rchk (#5273)
bind_cols() and bind_rows() reconstruct data frames too (#5275)
Sam Albers (1):
Add filter example to between (#5250)
Sara Stoudt (1):
New vignette: dplyr <-> base R (#4825)
Saras Windecker (1):
Remove dr_dplyr() (#4475)
Shane Piesik (1):
Add arrange() examples of tidy eval and across() (#4830)
Sina Rueeger (2):
Just a dot at the end of the sentence.
Moved the alternative up to right after the badge + changed proposed alternative `group_map()` to `group_modify()`.
Tobias Rockel (1):
Fix broken link to roxygen2 vignette (#5166)
Tobias Schieferdecker (1):
Document changed behaviour of 'right_join()' (#5257)
Will Beasley (1):
Remove underscore from `NULL_IF` (#4550)
Wim van der Ham (1):
Prevent slice_min/max from crashing when there are NA's (#4857)
arcoutte (1):
Fix README dtplyr link (#4553)
chayajones (1):
update examples to use starwars data, re-factor examples for clarity (#4827)
krterberg (1):
Tiny clarification for example for summarise_at (#4635)
takapai (1):
Document top_n() returns unsorted data frame (#4588)- Loading branch information
