tidyverse/dplyr

dplyr 1.1.0 revdep tracker

Closed this issue · 7 comments

DavisVaughan commented
dplyr 1.1.0 revdep tracker
hadley commented

I'll take these ones:

na_if() vctrs update #6329

  • 👍1
DavisVaughan commented

vctrs

na_if() vctrs update #6329

From join updates #5910

Assuming that most of these "produced warning" failures are from the new multi match join warning.

Misc

rename_with() now checks function output length

distinct() and sticky columns?

slice_*() strictness about n

slice() errors on named args

Set operations #6354

slice() with no args now drops all rows

From between updates #6260

arrange() radix ordering #6263

data_frame_() removal #6277

dplyr::tbl_sum() removal in favor of pillar::tbl_sum() #6284

case_when() uses vctrs #6300

  • epocakir: Uses case_when(~ ordered, TRUE ~ NA_integer_). Needs to use GFR_stages[NA_integer_] for default. alwinw/epocakir#41
  • huxtable: Tries to use a mix of matrix and vector logicals on the LHS. Should probably un-dim, apply case-when, then re-dim. hughjonesd/huxtable#234
  • unpivotr: case_when() was being used to return calls, but now requires vector types. It was also making this vctrs bug pop up, but was easily worked around r-lib/vctrs#1300 nacnudus/unpivotr#58

if_else() uses vctrs #6317

lead() and lag() update #6336

n_distinct() doesn't allow named args

Join check_dots_unused() (our fault) #6599

Join check dots their fault

Weird ranking functions missing links issue

I think CRAN will ignore this so I'm marking it complete.

  • heemod: Links to row_number() which previously was [dplyr:ranking]{dplyr::row_number()} but now row_number() is on a different help page

group_labels_details() weirdness? can't reproduce locally

These all seemed to be related to some old bugs in group_labels_details() that have disappeared on a rerun of the revdeps. I had my suspicions about them already because I couldn't reproduce locally, so i think they are fine.

  • GDPuc:
  • gmgm:
  • NHSRplotthedots:

False alarm?

  • SVMMaj: Looks like a bad tolerance issue in a test

Issues below this were walked back in dplyr itself, but we are tracking them here anyways.

Would be broken by nth(), first(), last() vctrs update if we made first(list(1)) use vec_slice() #6331

  • DataFakeR: %>% tidygraph::pull(table) %>% dplyr::first() on a list-col, expecting first() to give the first element
  • dyngen: Through GillespieSSA2, does str_split(" ") %>% first() expecting to get the first element of the list
  • dynwrap: Programmatic usage of first() on a list, expecting to get the first element
  • GillespieSSA2: does str_split(" ") %>% first() expecting to get the first element of the list (see also dyngen)
  • phenofit: last() on a list, expecting the last list element
  • TidyMultiqc: first() on a list expecting to get the first element
  • xpose: first() on a list expecting to get the first element

Would be broken by funs() deprecation #6389

  • AQuadtree: dplyr::summarise_at(qt2.act@data, sum.2, funs(sum))
  • bootnet: dplyr::summarize_each(., funs(fun(., digits = digits)))
  • caffsim: dplyr::summarise_at(., vars(value), funs(mean, sd, min, max))
  • carpenter: dplyr::mutate_all(., dplyr::funs(as.character))
  • clustrd: dplyr::summarise_all(., funs(mean))
  • coefplot: dplyr::mutate_at(...) some usage of funs in there
  • cometExactTest: dplyr::summarise_all(., funs(sum))
  • funModeling: dplyr::summarise_each_(., funs(group_func), vars_to_keep)
  • geosimilarity: dplyr::summarise_all(., funs(mean))
  • GFE: base::ifelse(type == "Jackknife", funs(varJack), funs(var))
  • glmdisc: dplyr::mutate_at(., dplyr::vars("X1", "X2"), dplyr::funs(as.integer))
  • graphicalVAR: dplyr::mutate_at(., funs(shift), .vars = vars)
  • IBCF.MTME: dplyr::mutate_if(., is.numeric, funs(round(., digits)))
  • interplot: in vignette
  • ipft: dplyr::summarise_all(., funs(mean))
  • Lahman: dplyr::summarise_each(., funs(sum))
  • missCompare: dplyr::summarise_all(., funs(mymean))
  • MRFcov: dplyr::summarise_all(., dplyr::funs(sd(.)))
  • MSiP: dplyr::summarise_each(., funs(max(.)))
  • ncappc: dplyr::mutate_all(., dplyr::funs(as.numeric(as.character(.))))
  • PAC: dplyr::summarise_all(., funs(sum))
  • PDtoolkit: dplyr::summarise_at(...)
  • phase1PRMD: dplyr::mutate_at(...)
  • PPforest: in vignette
  • prevtoinc: dplyr::mutate_all(., dplyr::funs(unlist))
  • prozor: dplyr::mutate_at(pepinfo, peptide, dplyr::funs(as.character))
  • psychonetrics: dplyr::summarise_at(., funs(mean(., na.rm = TRUE)), .vars = vars)
  • randomForestExplainer: dplyr::summarize_at(., vars, funs(max(., na.rm = TRUE)))
  • ruler: in vignette
  • sfc: mutate_each(funs(as.character(.)))
  • sjmisc: dplyr::summarise_all(...)
  • sjPlot: in vignette
  • skynet: dplyr::mutate_all(., funs(ifelse(is.na(.), 0, .)))
  • staRdom: in vignette
  • statVisual: dplyr::summarise_at(...)
  • tabularaster: in vignette
  • tidyboot: dplyr::summarise_at(., vars(mean), funs(ci_upper, mean, ci_lower))
  • useful: mutate_at(...)
  • xray: mutate_all(...)

Weird bind_rows() or bind_cols() missing links issue

  • blsR: Links to bind_rows() which previously was [dplyr:bind]{dplyr::bind_rows()} but now bind_rows() has its own help page
  • dtrackr:
  • ffp:
  • getLattes:
  • jstable:
  • mark:
  • mrgsim.parallel:
  • RKorAPClient:
  • salesforcer
  • sf:
  • sftime:
  • sjmisc:
  • skimr:
  • slackr:
  • soiltestcorr:
  • styler:
  • tidygraph:
  • tsibble:

Would be broken by slice_*() check_dots_empty() in generic

  • tidyterra: slice_sample.SpatRaster(.keep_extent = TRUE) extra argument counted as dots
lionel- commented

nth(), first(), last() vctrs update #6331

  • confoundr: "Can't convert default to match type of x ", first(x, default = "NA") with double x - jwjackson/confoundr#3
  • divseg: Does dplyr::first(dplyr::cur_data()) expecting to get the first col of the data frame? Should use pull(). - christopherkenny/divseg#1
  • finnts: dplyr::filter(Date==max(Date)) %>% dplyr::last() on a df where they should have used pull(-1) - microsoft/finnts#117
  • FRK: Not sure, possibly related but hard to diagnose - andrewzm/FRK#7
  • janitor: dplyr::last(dat) on a data frame expecting last column, should use pull() - sfirke/janitor#491
  • questionr: Broken because of janitor
  • tidygraph: dplyr::count() %>% dplyr::first() on a data frame, expecting first column, should use pull() - thomasp85/tidygraph#168
  • phenofit: eco-hydro/phenofit#14

select_vars() deprecation #6389

n_distinct() doesn't allow 0 args

Join check_dots_unused() (actual bug)

distinct() returns columns in requested order now #6156

Misc

  • starschemar: Looks like that have superclassed data.frame, something we don't support. x must be a vector, not a <tbl_df/tbl/data.frame/dimension_table> object - Sent mail with link to lionel-/starschemar@a392623

Unknown

  • r2dii.analysis: some calculation in tests is off, but very complex chain of calculations to get there - #6607
  • RNeXML: sort(meta.cont[, "Meta"]) not equal to sort(unique(meta.nested[, "meta"])) in test_meta_extract.R:204 - ropensci/RNeXML#254

na_if() vctrs update #6329

lionel- commented

New problems revealed in revdep checks of #6550:

New exported where() function causes an import conflict

Caused by #6609. where() is also exported by ape and imager packages.

DavisVaughan commented

From 24ea5ae

DavisVaughan commented

3 more detected only by CRAN's checks. We aren't sure how they slipped through ours.

hadley commented

On CRAN 🥳