doc: Updated math/R-emdbook to 1.3.12

(mef)

(math/R-compositions) Updated 2.0.2 to 2.0.4

Version 2.0-4

  * F77_ calling conventions adapted from fixed length strings to
    length parameter in BLAS calls
  * fitDirichlet returned loglikehood fixed (by adding a "-" in front
    of n)
  * as.matrix.rmult method returns now a matrix with the rmult class
    stripped; in this way, derivate functions of `lm()` work properly
    (e.g. `anova`, `summary.manova`, `MASS::lda`)

Version 2.0-3

  * kernel density estimation on the simplex with Dirichlet kernels
    (kdeDirichlet)

Version 2.0-2.9000
  * own implementation of dDirichlet

(mef)

doc: Updated math/R-emdbook to 1.3.12

(mef)

(math/emdbook) Updated 1.3.11 to 1.3.12

(CMD Rdconv -t txt math/R-emdbook/work/emdbook/inst/NEWS.Rd |head -30)
Changes in version 1.3.12:

        * documentation/CRAN-compatibility update

(mef)

doc: Updated math/R-e1071 to 1.7.12

(mef)

(math/R-e1071) Updated 1.7.9 to 1.7.12

(R CMD Rdconv -t txt math/R-e1071/work/e1071/inst/NEWS.Rd |head -20)

Changes in version 1.7-12:

        * add 'tune.gknn()' wrapper

Changes in version 1.7-11:

        * cosmetic changes

Changes in version 1.7-10:

        * small fix in 'coef.svm' for sparse data

(mef)

doc: Updated math/R-date to 1.2.40

(mef)

(math/R-date) Updated 1.2.39 to 1.2.40

2022-10-07  Kurt Hornik  <Kurt.Hornik@wu.ac.at>

* DESCRIPTION: New version is 1.2-40.

* src/char_date.c: Add missing prototypes.

(mef)

doc: Updated math/R-classInt to 0.4.8

(mef)

(math/R-classInt) Updated 0.4.3 to 0.4.8

## Version 0.4-8

- #18 and #38: `classIntervals()` has a new style `"box"`, where a box
  map is an augmented quartile map, with an additional lower and
  upper category. When there are lower outliers, then the starting
  point for the breaks is the minimum value, and the second break is
  the lower fence. When there are no lower outliers, then the
  starting point for the breaks will be the lower fence, and the
  second break is the minimum value (there will be no observations
  that fall in the interval between the lower fence and the minimum
  value) (@angela-li, @dieghernan).

## Version 0.4-7

- A new helper function `classify_intervals()` is introduced to return
  a vector of class intervals of same length as input (@JosiahParry)

- `classIntervals()` has a new style `"maximum"` which returns maximum
  breaks classification based on the pysal library
  [mapclassify](https://pysal.org/mapclassify/index.html)
  (@JosiahParry)

- `findCols()` now takes new argument `factor` which, when `TRUE`
  returns class membership as a factor with intervals as labels
  (@JosiahParry)

(mef)

Updated sysutils/py-subprocess-tee, time/py-cftime, devel/py-cbor2, devel/py-isort

(adam)

doc: Updated math/R-circular to 0.4.95

(mef)

(math/R-circular) Updated 0.4.93 to 0.4.95, no explicit NEWS.md

(mef)

doc: Updated math/R-chron to 2.3.58

(mef)

(math/R-chron) Updated  2.3.56 to 2.3.58

2022-10-04  Kurt Hornik  <Kurt.Hornik@wu.ac.at>

* DESCRIPTION: New version is 2.3-58.

* R/chron.R (as.chron.POSIXt): Update for recent
as.character.POSIXt() changes in base R.

2022-05-30  Kurt Hornik  <Kurt.Hornik@wu.ac.at>

* DESCRIPTION: New version is 2.3-57.

* src/Rchron.h: Add typedef for legacy 'Sint' type.

2021-03-08  Kurt Hornik  <Kurt.Hornik@wu.ac.at>

* man/chron.Rd:
* man/dates.Rd:
* man/format.chron.Rd:
* man/hours.Rd:
* man/seq.dates.Rd:
Avoid partial matching.

(mef)

doc: Updated math/R-broom to 1.0.2

(mef)

(math/R-broom)  Updated 0.8.0 to 1.0.2

# broom 1.0.2

* The default `data` argument for `augment.coxph()` and
  `augment.survreg()` has been transitioned from `NULL` to
  `model.frame(x)` (#1126 by `@capnrefsmmat`).
* Migrated 'ggplot2' from strong to weak dependency, i.e. moved from
  `Imports` to `Suggests`.
* Fixed a bug where `augment()` results would not include residuals
  when the response term included a function call (#1121, #946, #937,
  #124).

# broom 1.0.1

* Improves performance of `tidy.lm()` and `tidy.glm()` for full-rank
  fits (#1112 by `@capnrefsmmat`).

* Moves forward with deprecation of tidiers for sparse matrices
  outputted from the Matrix package, initially soft-deprecated in
  broom 0.5.0. The Matrix tidiers were light wrappers around coercion
  methods that will now be deprecated from Matrix itself in the
  upcoming 1.4-2 release. The affected methods are
  `tidy.sparseMatrix()`, `tidy.dgCMatrix()`, and
  `tidy.dgTMatrix()`. Note that `tidy.confusionMatrix()`, for relevant
  objects outputted from the caret package, is unaffected (#1113).

* `tidy.anova()` works again with `anova` objects from the `lme4`
  package (broken by addition of the `terms` column in the previous
  release)

# broom 1.0.0

broom 1.0.0 is the first "production" release of the broom package,
and includes a number of notable changes to both functionality and
governance.

As of this release, the broom team will be following a set of
guidelines that clarify the scope of further development on the
package. Given the package's wide use and long history, these
guidelines _prioritize backward compatibility_ over internal
consistency and completeness. You can read those guidelines
[here](https://broom.tidymodels.org/articles/)!

We've also made notable changes to error handling in this release:

* Adds minimal ellipsis checking to warn on commonly misspecified
  arguments passed through ellipses. Notably:
    + `tidy()` methods will now warn when supplied an `exponentiate`
      argument if it will be ignored.
    + `augment()` methods will now warn when supplied a `newdata`
      argument if it will be ignored.

* The warning regarding tidiers only maintained via dispatch to `lm`
  and `glm` is now displayed only once per session, per unique
  dispatch. That is, if a `class_a` object is tidied using a `(g)lm`
  method, broom will not warn when tidying `class_a` objects for the
  rest of the session, but if a `class_b` object is tidied using a
  `(g)lm` method in the same session, broom will warn again (#1101).

Other fixes and improvements:

* Add `exponentiate` argument to `tidy.boot()` (#1039).
* Update in `tidy.htest()` converting matrix-columns to vector-columns (#1081).
* Address failures in `tidy.glht()` with `conf.int = TRUE` (#1103).
* Address failures in `tidy.zoo()` when input data does not have `colnames`
  (#1080).
* Transition tidiers for bivariate linear or spline-based interpolation---using
  list tidiers to interface with objects from the akima package is now
  considered off-label. See the interp package for a FOSS alternative.
* Address failures in `tidy.svyolr()` when `p.values = TRUE`. Instead of aliasing
  `tidy.polr()` directly, `tidy.svyolr()` lightly wraps that method and
  warns if `p.values` is supplied (#1107).
* Adds a `term` column and introduces support for `car::lht()` output in
  `tidy.anova()` (#1106 by `@grantmcdermott`).
* Adds a dedicated `glance.anova` method (which previously dispatched to the
  deprecated `glance.data.frame()` tidier, #1106 by `@grantmcdermott`).

(mef)

py-isort: updated to 5.11.3

5.11.3 December 16 2022
- Fixed 2007: settings for py3.11
- Fixed 2038: packaging pypoetry
- Docs: renable portray
- Ci: add minimum GitHub token permissions for workflows
- Ci: general CI improvements
- Ci: add release workflow

5.11.1 December 12 2022
- Hotfix 2031: only call `colorama.init` if `colorama` is available

5.11.0 December 12 2022
- Added official support for Python 3.11
- Dropped support for Python 3.6
- Fixed problematic tests
- Fixed 1960: Rich compatibility
- Fixed 1945, 1986: Python 4.0 upper bound dependency resolving issues @staticdev
- Fixed Pyodide CDN URL
- Docs: clarify description of use_parentheses
- Fixed 1976: `black` compatibility for `.pyi` files @XuehaiPan
- Implemented 1683: magic trailing comma option
- Add missing space in unrecoverable exception message
- Fixed 1895: skip-gitignore: use allow list, not deny list @bmalehorn
- Fixed 1917: infinite loop for unmatched parenthesis
- Docs: shared profiles
- Fixed build-backend values in the example plugins
- Remove reference to jamescurtin/isort-action
- Split long cython import lines
- Update plone profile: copy of `black`, plus three settings.
- Fixed 1815, 1862: Add a command-line flag to sort all re-exports
- Fixed 1854: `lines_before_imports` appending lines after comments
- Remove redundant `multi_line_output = 3` from "Compatibility with black"
- Add tox config example
- Docs: add examples for frozenset and tuple settings
- Docs: add multiple config documentation

(adam)

doc: Updated math/R-car to 3.1.1

(mef)

(math/R-car) Updated 3.0.13 to 3.1.1

Changes to Version 3.1-1

  o Fixed bug in hccm(): when some cases are fit exactly all versions
    of the hccm estimate are singular and inconsistent.  A error message
    is now returned.

  o Fixed bug in Anova.survreg() for Wald tests (reported by Megan
    Taylor Jones), which failed with a spurious "missing residual df"
    error.

  o Made Anova.lm() more robust when there are aliased coefficients
    (following report of cryptic output by Taiwo Fagbohungbe).

  o Added reg.function, reg.function.col, and mouseMode arguments to
    scatter3d(), and mouseMode to crPlot3d().

  o Added fill and associated arguments to influencePlot() (initial
    implementation of Michael Friendly).

  o Added confidenceEllipses() function to create matrix of all
    pairwise confidence ellipses for a regression model (suggestion of
    Michael Friendly).

  o Moved the pointLabel() function from the maptools package in
    anticipation of the retirement of that package, and with the
    permission of the function author, Tom Short.

Changes to Version 3.1-0

  o New 3D added-variable and component+residual plots, avPlot3d() and
    crPlot3d().

  o More flexible recode()/Recode().

  o Added Anova.svycoxph() to prevent selection of test.statistic="LR".

  o New take on vif.lm().

(mef)

py-cbor2: updated to 5.4.6

5.4.6** (2022-12-07)
- Fix MemoryError when decoding Tags on 32bit architecture. (Sekenre)

**5.4.5** (2022-11-29)
- Added official Python 3.11 support (agronholm)
- Raise proper exception on invalid bignums (Øyvind Rønningstad)
- Make Tagged item usable as a map key (Niels Mündler)
- Eliminate potential memory leak in tag handling (Niels Mündler)
- Documentation tweaks

(adam)

py-cftime: updated to 1.6.2

version 1.6.2 (release tag v1.6.2rel)
=====================================
* num2date should not fail on an empty integer array.
* longdouble keyword in date2num so that a roundtrip from a time to a date
  and back again does not lose microsecond precision when the units require
  the times be encoded as floating point values
* added strptime method.
* cibuildwheel wheel-building workflow added to github actions by @ocefpaf (triggers binary
  wheel builds and uploads to pypi automatically when GH release created).

version 1.6.1 (release tag v1.6.1rel)
=====================================
* fix failing tests on windows with numpy 1.23.0
* expose to_tuple module function in public API.

version 1.6.0 (release tag v1.6.0rel)
=====================================
* fix for masked array inputs.
* improved performance of the num2date algorithm, in some cases providing
  an over 100x speedup.
* fix for date2index for select != 'exact' when select='exact' works

(adam)

py-subprocess-tee: updated to 0.4.1

v0.4.1

Bugfixes

Correct trove metadata related to supported platforms
Move mypy config to pyproject.toml
Stop using deprecated get_event_loop()
Configure packit
Add "setuptools >= 61.0" to build system requires
Clear build dependencies

(adam)

Updated devel/blosc, archivers/unrar, time/py-pytz, devel/py-pylint

(adam)

py-pylint: updated to 2.15.9

What's new in Pylint 2.15.9?

False Positives Fixed

Fix false-positive for used-before-assignment in pattern matching with a guard.

Other Bug Fixes

Pylint will no longer deadlock if a parallel job is killed but fail immediately instead.
When pylint exit due to bad arguments being provided the exit code will now be the expected 32.
Fixes a ModuleNotFound exception when running pylint on a Django project with the pylint_django plugin enabled.

(adam)

py-pytz: updated to 2022.7

2022.7
Unknown changes

(adam)

unrar: updated to 6.2.3

6.2.3
Unknown changes

(adam)

blosc: updated to 1.21.3

Changes from 1.21.2 to 1.21.3

Internal LZ4 codec updated to 1.9.4.
Internal BloscLZ codec updated to 2.5.1.

(adam)

(graphics/R-tkrplot) Added version 0.0.27

(mef)

(graphics/R-tkrplot) import R-tkrplot-0.0.27

Simple mechanism for placing R graphics in a Tk widget.

(mef)

doc: Updated math/R-akima to 0.6.3.4

(mef)

(math/R-akima) Updated 0.6.3.4, explict ChangeLog unknown

(mef)

Fix uncommitted package linked, tks jperkin@

(mef)

(benchmarks/R-profvis) (www/R-downlit) Added

(mef)

(www/R-downlit) import R-downlit-0.4.0

Syntax highlighting of R code, specifically designed for the needs of
'RMarkdown' packages like 'pkgdown', 'hugodown', and 'bookdown'. It
includes linking of function calls to their documentation on the web,
and automatic translation of ANSI escapes in output to the equivalent
HTML.

(mef)

(benchmarks/R-profvis) import R-profvis-0.3.7

Interactive visualizations for profiling R code.

(mef)

doc: Updated math/R-RcppEigen to 0.3.3.9.3

(mef)

(math/R-RcppEigen) Updated  0.3.3.9.1 to 0.3.3.9.3

2022-11-04  Dirk Eddelbuettel  <edd@debian.org>

* src/init.c: Add 'void' for proper prototype pleasing clang-15

* R/fastLm.R (summary.fastLm,print.fastLm): Refer to correct and full
variable name df.residual in the returned object

* .github/workflows/ci.yaml (jobs): Update to actions/checkout@v3

2022-09-15  Jonah Gabry  <jgabry@gmail.com>

* inst/skeleton/rcppeigen_hello_world.cpp: Correct typo

2022-04-08  Dirk Eddelbuettel  <edd@debian.org>

* DESCRIPTION (Depends): Add a versioned dependency on R 3.6.0 or
later because of our use of FCONE to support USE_FC_LEN_T

2022-04-05  Dirk Eddelbuettel  <edd@debian.org>

* DESCRIPTION (Version, Date): CRAN Release 0.3.3.9.2
* inst/NEWS.Rd: Release 0.3.3.9.2

2022-04-04  Dirk Eddelbuettel  <edd@debian.org>

* src/fastLm.cpp: Add FCONE in two calls for improved Fortran and C
character interface per Writing R Extensions, Section 6.6.1

2022-01-16  Dirk Eddelbuettel  <edd@debian.org>

* inst/tinytest/test_wrap.R: Added (optional) large memory wrap tests
* inst/tinytest/cpp/wrap.cpp: Added C++ part of test

* .codecov.yml: Added to not trigger PR fail for small additions

2022-01-16  Mikael Jagan  <jaganmn@mcmaster.ca>

* inst/include/RcppEigenWrap.h: Refine use plain dense wrap() change

2022-01-15  Mikael Jagan  <jaganmn@mcmaster.ca>

* inst/include/RcppEigenWrap.h: Use R_xlen_t for vectors rows + cols

2021-12-29  Dirk Eddelbuettel  <edd@debian.org>

* README.md: Add total downloads badge

2021-12-08  Dirk Eddelbuettel  <edd@debian.org>

* README.md: Remove unused continuous integration artifact and badge

2021-10-13  Dirk Eddelbuettel  <edd@debian.org>

* inst/CITATION: Refinment of doi use

2021-10-10  Dirk Eddelbuettel  <edd@debian.org>

* inst/CITATION: Switch JSS url to doi form per JSS request
* README.md: Idem

2021-07-19  Dirk Eddelbuettel  <edd@debian.org>

* inst/include/RcppEigenWrap.h: Two more #nocov tags
* src/fastLm.cpp: One more #nocov tag

2021-07-18  Dirk Eddelbuettel  <edd@debian.org>

* inst/tinytest/test_fastLm.R: Add tests for summary
* src/fastLm.cpp: Add a few #nocov tags

2021-07-17  Dirk Eddelbuettel  <edd@debian.org>

* DESCRIPTION (Date, Version): Roll minor version

* R/RcppEigen.package.skeleton.R (RcppEigen.package.skeleton): Also
import RcppEigen in DESCRIPTION

2021-06-07  Dirk Eddelbuettel  <edd@debian.org>

* DESCRIPTION (Date, Version): Roll minor version

* inst/tinytest/test_misc.R: New test file
* inst/tinytest/test_fastLm.R: Added tests

* R/fastLm.R: Add single one-line nocov tag

* .Rbuildignore: Add .covignore and .codecov.yml

2021-06-06  Dirk Eddelbuettel  <edd@debian.org>

* README.md: Added coverage badge

* .github/workflows/ci.yaml (jobs): Turn on coverage
* .covrignore: Added
* .codecov.yml (ignore): Idem

* R/RcppEigen.package.skeleton.R: Set nocov

2021-05-09  Dirk Eddelbuettel  <edd@debian.org>

* DESCRIPTION (URL): Add GitHub repo to URL field

2021-01-02  Dirk Eddelbuettel  <edd@debian.org>

* R/RcppEigen.package.skeleton.R: Wrap any() around grepl()

2020-12-25  Dirk Eddelbuettel  <edd@debian.org>

* .github/workflows/ci.yaml: Small tweaks to CI YAML file

(mef)

Make it build on NetBSD/macppc

(martin)

R-*: Temporarily comment out dependencies.

These dependencies haven't been committed yet, unbreak bulk builds until they
can be added.

(jperkin)

doc: Updated math/R-RPostgreSQL to 0.7.4

(mef)

(math/R-RPostgreSQL) Updated 0.7.3 to 0.7.4, explict NEWS.md unknown

(mef)

doc: Updated math/R-RNetCDF to 2.6.1

(mef)

(math/R-RNetCDF) Updated 2.5.2 to 2.6.1

Version 2.6-1, 2021-06-25
  * Windows: update binary packages to netcdf 4.9.0 with OpenDAP

(mef)

doc: Updated math/R-RColorBrewer to 1.1.3

(mef)

(math/R-RColorBrewer) Updated 1.1.2 to 1.1.3, NEWS.md unknown

(mef)

doc: Updated math/R-R2HTML to 2.3.3

(mef)

(math/R-R2HTML) Updated 2.3.2 to 2.3.3

# 2022-05-22 - Version 2.3.3
- Fix HTML.ftable failing when there are no column variables
- Fix returning results, invisibly

(mef)

doc: Updated math/R-ISLR to 1.4

(mef)

(math/R-ISLR) Updated 1.2 to 1.4, NEWS.md unknown

(mef)

doc: Updated math/R-FNN to 1.1.3.1

(mef)

(math/R-FNN) Updated 1.1.3 to 1.1.3.1, ChangeLog not updated

(mef)

doc: Updated math/R-DEoptimR to 1.0.11

(mef)

(math/R-DEoptimR) Updated 1.0.9 to 1.0.11, NEWS not updated

(mef)

doc: Updated math/R-DBI to 1.1.3

(mef)

(math/R-DBI) Updated 1.1.1 to 1.1.3

# DBI 1.1.3 (2022-06-18)

## Features
- `dbAppendTable()` accepts `Id` (#381, @renkun-ken).

## Documentation
- `?dbSendQuery` and related methods gain a section "The data
  retrieval flow" (#386).

- `?dbSendStatement` and related methods gain a section "The command
  execution flow" (#386).

# DBI 1.1.2 (2021-12-19)

## Features
- Use `dbQuoteLiteral()` in default method for `sqlData()` (#362, #371).
- Update specification with changes from DBItest 1.7.2 (#367).

## Documentation
- The pkgdown documentation for DBI generics (e.g. `?dbConnect`)
  contains clickable links to all known backends (except ROracle), and
  an explanatory sentence (#360).

- `?Id` gains better examples (#295, #370).

- Elaborate on status of `dbWriteTable()` in the documentation (#352,
  #372).

- Make method definition more similar to S3. All `setMethod()` calls
  refer to top-level functions (#368).

- `?dbReadTable` and other pages gain pointers to `Id()` and `SQL()` (#359).

(mef)

doc: Updated math/R-DAAG to 1.25.4

(mef)

(math/R-DAAG) Updated 1.24 to 1.25.4

Changes in DAAG version 1.25.1-3
  o A number of URLs have been updated.
  o The bomregions2021 dataset updates the bomregions data
    to 2021.
  o The dataset mdbAVtJtoD, with monthly Murray-Darling
    basin data from January 1950 through to February 2022,
    has been added.

Changes in DAAG version 1.25

UPDATES
  o Vignettes sources are now R Markdown, with the
    format bookdown::html_document2 format combined
    with base_format: rmarkdown::html_vignette
    (This combination allows automatic numbering of
    figures, and cross-references.)
  o The `bomregions2018` dataset updates `bomregions`
    to include data through to 2018.  The updates
    `bomregions2012` and `bomregions2015` have been
    omitted.  A further column `avDMI` has been added,
    holding values (from 1960 onwards) of the Annual
    average (Indian) Dipole Mode Index.
  o Web links on what is now the `bomregions2018` help
    page have been updated.  The url in the DESCRIPTION
    file now links to the development version of DAAG
    on the gitlab site.
  o The `head.injury` dataset (which was a copy of the
    `headInjury` dataset) has been removed.
  o Help pages for related data sets have, in several
    cases, been combined.

                Changes in DAAG version 1.24-1

NEW FEATURES
  o The dataset coralPval, holding 3072 p-values from a gene
    expression array comparison between post- and pre-settlement
    coral larvae, has been added.
  o The dataset bomregions2015 (updates bomregions to include data
    up to 2015) has been added.
  o The dataset lognihills, which replaces values in all 4 columns
    of nihills by log(values), has been added. Column names are
    logdist, logclimb, logtime and logtimef.
  o The generic function sumry(), with a method and print method
    for glm (but not other) objects, has been added.
  o The function DAAGtheme(), used to generate themes for use in
    an intended 4th edition of Data Analysis and Graphics Using R,
    has been added.
  o The functions CVlm() (alias cv.lm()), simulateLinear(),
    qreference() and sampdist() all now return lattice objects.
    If the result is returned to the command line, the graph is
    plotted just as before.

BUG FIXES
  o A line that subtracted 1 from the bottom and top margins,
    causing an error if the margin was already set to < 1, was
    removed from onetPermutation(), twot() and twotPermutation().
  o Code has been amended so that term labels ('term.labels') are
    given by attr(obj$terms, "term.labels").  The code now
    replaced expected to find 'term.labels' among the attributes
    of the summary object; this would have failed starting
    several R releases back.

(mef)

(textproc/R-rbibutils) (textproc/R-Rdpack) Added.

(mef)

(textproc/R-Rdpack) import R-Rdpack-2.4

Functions for manipulation of R documentation objects, including
functions reprompt() and ereprompt() for updating 'Rd' documentation
for functions, methods and classes; 'Rd' macros for citations and
import of references from 'bibtex' files for use in 'Rd' files and
'roxygen2' comments; 'Rd' macros for evaluating and inserting snippets
of 'R' code and the results of its evaluation or creating graphics on
the fly; and many functions for manipulation of references and Rd
files.

(mef)

(textproc/R-rbibutils) import R-rbibutils-2.2.11

Read and write 'Bibtex' files. Convert between bibliography formats,
including 'Bibtex', 'Biblatex', 'PubMed', 'Endnote', and 'Bibentry'.
Includes a port of the 'bibutils' utilities by Chris Putnam
<https://sourceforge.net/projects/bibutils/>. Supports all
bibliography formats and character encodings implemented in
'bibutils'.

(mef)

doc: Updated math/R-dplyr to 1.0.10

(mef)

(math/R-dplyr) Updated 1.0.9 to 1.0.10

# dplyr 1.0.10

Hot patch release to resolve R CMD check failures.

(mef)

doc: Updated devel/R-IRkernel to 1.3.1

(mef)

(devel/R-IRkernel) Updated 1.0.2 to 1.3.1, explicit NEWS.md unknown

(mef)

doc: Updated devel/R-IRdisplay to 1.1

(mef)

(devel/R-IRdisplay) Updated 1.0 to 1.1, explicit NEWS.md unknown

(mef)

doc: Updated devel/R-BH to 1.78.0.0

(mef)

(devel/R-BH) Updated 1.75.0.0 to 1.78.0.0

2021-12-14  Dirk Eddelbuettel  <edd@debian.org>

* DESCRIPTION (Version, Date): Release 1.78.0-0

2021-12-13  Dirk Eddelbuettel  <edd@debian.org>

* DESCRIPTION (Version, Date): See Version to 1.77.99-0 for interim
test release

* inst/include/boost: As before, adjusted to CRAN Policy requirements
of i) not path names under 104 chars and b) no diagnostic suppression
* local/patches/bh_1.78.0-0*patch: Document these

2021-12-12  Dirk Eddelbuettel  <edd@debian.org>

* inst/include/boost/*: Upgraded to Boost 1.78

* local/scripts/CreateBoost.sh: Update for 1.78.0, added lambda2

2021-12-11  Dirk Eddelbuettel  <edd@debian.org>

* README.md: Add total download badge, label other as monthly

2021-11-04  Dirk Eddelbuettel  <edd@debian.org>

* README.md: Remove Travis badge
* .travis.yml: Remove Travis YAML config

2021-03-19  Dirk Eddelbuettel  <edd@debian.org>

* DESCRIPTION (URL): List GitHub repo and package page

(mef)

doc: Updated devel/R-xfun to 0.35

(mef)

(devel/R-xfun) Updated 0.31 to 0.35

# CHANGES IN xfun VERSION 0.35

- Added a new argument `token` to `protect_math()` to optionally
  include a token around math expressions.

- `base64_uri()` relies less on the **mime** package now. For some
  common file extensions (e.g., `.jpg`/`.png`), this function knows
  their MIME types.

- `stringsAsStrings()` has been removed from this package.

# CHANGES IN xfun VERSION 0.34

- Added a new function `alnum_id()` to generate ID strings from a character vector.

- The function `stringsAsStrings()` has been deprecated.

# CHANGES IN xfun VERSION 0.33

- Reverted the change for #68: the characters `-+!_#` are no longer
  accepted by default in filename extensions, since they are
  relatively rare and caused a breakage in rstudio/bookdown#1369. If
  you wish to allow for these characters, you may use the new `extra`
  argument in `file_ext()` and related functions, e.g.,
  `xfun::file_ext(x, extra = '-+!_#')`.

- The function `stringsAsStrings()` will be deprecated in a future
  release of **xfun**, because the global option `stringsAsFactors =
  FALSE` has become the default in base R since 4.0.0.

# CHANGES IN xfun VERSION 0.32

- Added a function `shrink_images()` to shrink images to a maximum
  width using the **magick** package (thanks, @apreshill,
  rstudio/blogdown#614).

- Added a function `tinify_dir()` as a wrapper of `tinify()` to
  compress images under a directory.

- `file_ext()` supports more file extensions now, such as `.c++`,
  `.FB2K-COMPONENT`, and so on (thanks, @tentacles-from-outer-space,
  #68).

- Fixed the issue that `xfun::base_pkgs()` could hang R (thanks,
  @mmaechler, #66).

- The `...` argument in `dir_create()` was not passed to `dir.create()`.

(mef)

doc: Updated devel/R-whisker to 0.4.1

(mef)

(devel/R-whisker) Updated 0.4 to 0.4.1 explicit NEWS.md unknown

(mef)

doc: Updated devel/R-tidyselect to 1.2.0

(mef)

(devel/R-tidyselect) Updated 1.1.2 to 1.2.0

# tidyselect 1.2.0

## New features

* New `tidyselect_data_proxy()` and `tidyselect_data_has_predicates()`
  allows tidyselect to work with custom input types (#242).

* New `eval_relocate()` for moving a selection. This powers `dplyr::relocate()`
  (#232).

## Lifecycle changes

* Using `all_of()` outside of a tidyselect context is now deprecated (#269).
  In the future it will error to be consistent with `any_of()`.

* Use of `.data` in tidyselect expressions is now deprecated to more cleanly
  separate tidy-select from data-masking. Replace `.data$x` with `"x"` and
  `.data[[var]]` with `any_of(var)` or `all_of(var)` (#169).

* Use of bare predicates (not wrapped in `where()`) and indirection (without
  using `all_of()`) have been formally deprecated (#317).

## Minor improvements and bug fixes

* Selection language:

  * `any_of()` generates a more informative error if you supply too many
    arguments (#241).

  * `all_of()` (like `any_of()`) returns an integer vector to make it easier
    to combine in functions (#270, #294). It also fails when it can't find
    variables even when `strict = FALSE`.

  * `matches()` recognises and correctly uses stringr pattern objects
    (`stringr::regex()`, `stringr::fixed()`, etc) (#238). It also now
    works with named vectors (#250).

  * `num_range()` gains a `suffix` argument (#229).

  * `where()` is now exported, like all other select helpers (#201),
    and gives more informative errors (#236).

* `eval_select()` with `include` now preserves the order of the variables
  if they're present in the selection (#224).

* `eval_select()` always returns a named vector, even when renaming is not
  permitted (#220).

* `eval_select()` and `eval_relocate()` gain new `allow_empty` argument which
  makes it possible to forbid empty selections with `allow_empty = FALSE` (#252).

* `eval_select(allow_rename = FALSE)` no longer fails with empty
  selections (#221, @eutwt) or with predicate functions (#225). It now properly
  fails with partial renaming (#305).

* `peek_var()` error now generates hyperlink to docs with recent RStudio (#289).

* `vars_pull()` generates more informative error messages (#234, #258, #318)
  and gains `error_call` and `error_arg` arguments.

* Errors produced by tidyselect should now be more informative. Evaluation
  errors are now chained, with the child error call is set to the `error_call`
  argument of `eval_select()` and `eval_rename()`. We've also improved
  backtraces of base errors, and done better at propagating the root
  `error_call` to vctrs input checkers.

* `tidyselect_verbosity` is no longer used; deprecation messaging is now
  controlled by `lifecycle_verbosity` like all other packages (#317).

(mef)

doc: Updated devel/R-rversions to 2.1.2

(mef)

(devel/R-rversions) Updated 2.1.1 to 2.1.2

# rversions 2.1.2

No user visible changes.

(mef)

doc: Updated devel/R-roxygen2 to 7.2.3

(mef)

(devel/R-roxygen2) Updated 7.1.2 to 7.2.3

# roxygen2 7.2.3

* roxygen2 now supports HTML blocks in markdown. They are only included
  in the HTML manual. They can also be produced as the output of code
  chunks.

* Improved support for RStudio IDE.

# roxygen2 7.2.2

* `@includeRmd` calls `local_reproducible_output()` to make code run in
  included `.Rmd`s more consistent with other sources (#1431).

* Fix duplicated argument in `roxy_block()` to avoid CRAN removal.

# roxygen2 7.2.1

## Tags

* All built-in tags are now documented so that you can do (e.g.) `?"@param"`
  to get a basic description of `@param` and a pointer where to learn more
  (#1165). This is powered by a new `tags_list()` lists all tags defined by
  roxygen2 and `tags_metadata()` provides some useful information about them
  for use by (e.g.) IDEs (#1375).

* `@describeIn` can now be used to combine more types of functions
  (generics, methods and other functions) into a single topic.
  The resulting section organises the functions by type (#1181)
  and displays methods like function calls. Methods are recognized only if
  they extend the generic in the destination,or if the destination can
  heuristically be identified as a constructor.

* Code evaluated in inline markdown code chunks and `@eval`/`@evalRd`/
  `@evalNamespace` is now evaluated in an environment designed to be more
  reproducible and to suppress output that won't work in Rd (e.g. turning
  off colour and unicode support in cli) (#1351). They now also set
  knitr options `comment = #>` (#1380) and `collapse = TRUE` (#1376).

* `@export` will now export both the class and constructor function when
  applied to expressions like `foo <- setClass("foo")` (#1216).

* `@includeRmd` now gives better feedback when it fails (#1089).

## (R)markdown

* New `knitr_chunk_options` option (in the `Roxygen` entry of
  `DESCRIPTION` or in `man/roxygen/meta.R`) is added to the knitr chunk
  options that roxygen2 uses for markdown code blocks and inline
  code (#1390).

* PDF figures are only included the PDF manual, and SVG figures are only
  included in the HTML manual (#1399).

* You can now use alternative knitr engines in markdown code blocks (#1149).

* Generated HTML for code blocks never includes "NA" for language (#1251).

* Using a level 1 heading in the wrong tag now gives a more useful warning
  (#1374).

* Fix bug interpolating the results of indented inline RMarkdown (#1353).

## Other

* If you have a daily build of RStudio, the lists of changed Rd files are
  now clickable so you can immediately see the rendered development
  documentation (#1354).

* R6 documentation no longer shows inherited methods if there aren't any
  (#1371), and only links to superclass docs if they're actually available
  (#1236).

* Automated usage no longer mangles nbsp in default arguments (#1342).

# roxygen2 7.2.0

## New features

* The NAMESPACE roclet now preserves all existing non-import directives during
  it's first pre-processing pass. This eliminates the "NAMESPACE has changed"
  messages and reduces the incidence of namespace borking (#1254).

* `@inheritParams` now only inherits exact multiparameter matches, so if you're
  inheriting from a function with `@param x,y` you'll only get the parameter
  documentation if your function needs docs for both x and y (#950).

* All warning messages have been reviewed to be more informative and
  actionable (#1317). `@title` now checks for multiple paragraphs.
  `@export` gives a more informative warning if it contains too many lines.
  (#1074). All tags warn now if only provide whitespace (#1228), and
  problems with the first tag in each block are reported with the correct line
  number (#1235).

* If you have a daily build of RStudio, roxygen2 warnings will now include a
  clickable hyperlink that will take you directly to the problem (#1323).
  This technology is under active development across the IDE and the cli
  package but is extremely exciting.

## Minor improvements and bug fixes

* roxygen2 can once again read UTF-8 paths on windows (#1277).

* `@author`s are de-duplicated in merged documentation (@DanChaltiel, #1333).

* `@exportS3method pkg::generic` now works when `pkg::generic` isn't
  imported by your package (#1085).

* `@includeRmd` is now adapted to change in rmarkdown 2.12 regarding math
  support in `github_document()` (#1304).

* `@inherit` and friends perform less aggressive link tweaking, eliminating
  many spurious warnings. Additionally, when you do get a warning, you'll
  now always learn which topic it's coming from (#1135). Inherited
  `\ifelse{}{}{}` tags are now inserted correctly (without additional `{}`)
  (#1062).

* `@inherit` now supports inheriting "Notes" with `@inherit pkg::fun note`
  (@pat-s, #1218)

* Automatic `@usage` now correctly wraps arguments containing syntactically
  significant whitespace (e.g anonymous functions) (#1281) and non-syntactic
  values surrounded by backticks (#1257).

* Markdown:

    * Code blocks are always wrapped in `<div class="sourceCode">`
      even if the language is unknown (#1234).

    * Links with markup (e.g. ``[foo `bar`][target]``) now cause an informative
      warning instead of generating invalid Rd.

    * Curly braces in links are now escaped (#1259).

    * Inline R code is now powered by knitr. Where available, (knit) print
      methods are applied (#1179). This change alters outputs and brings roxygen
      in line with console and R markdown behavior. `x <- "foo"` no longer
      inserts anything into the resulting documentation, but `x <- "foo"; x`
      will. This also means that returning a character vector will insert
      commas between components, not newlines.

* roxygen2 no longer generates invalid HTML (#1290).

* DOIs, arXiv links, and urls in the `Description` field of the `DESCRIPTION`
  are now converted to the appropriate Rd markup (@dieghernan, #1265, #1164).
  DOIs in the `URL` field of the `DESCRIPTION` are now converted to Rd's
  special `\doi{}` tag (@ThierryO, #1296).

(mef)

doc: Updated textproc/R-rmarkdown to 2.19

(mef)

(textproc/R-rmarkdown) Updated 2.10 to 2.19

rmarkdown 2.19
================================================================================
- No longer use `xfun::isFALSE()` internally.

rmarkdown 2.18
================================================================================
- `rmarkdown::render()` argument `output_file` is now used for all
  output formats, if there is only one name for multiple output
  formats (thanks, @MaelAstruc, #2421).

- Custom output formats can provide a custom function to convert
  Markdown input to the desired output format now. To do so, define
  the output format via `output_format(pandoc =
  pandoc_options(convert_fun = function(...) {}))`. The default
  converting function is `pandoc_convert()` in this package. See the
  help page `?pandoc_options` for more details.

rmarkdown 2.17
================================================================================
- `rmarkdown::render_site()` will now also look for `index.rmd` in
  addition to `index.Rmd` for custom site generator in YAML with
  `site:` (thanks, @kamalsacranie, #2409).

- Icons in tabset dropdown are now correctly shown, including when
  `self_contained: false` (thanks, @cpsievert, #1697).

rmarkdown 2.16
================================================================================
- Pandoc 2.19 has deprecated the argument `--self-contained`. If you
  have installed Pandoc 2.19+, **rmarkdown** will use
  `--embed-resources --standalone` as recommended by Pandoc for output
  formats that use the option `self_contained = TRUE` (#2382).

- `pandoc_version()` returns a version number of the form
  `X.Y.Z.YYYY.MM.DD` for the nightly build of Pandoc now, where
  `YYYY.MM.DD` is its build date. Previously it would return
  `X.Y.Z.9999` to indicate the nightly version.

rmarkdown 2.15
================================================================================
- Fix navbar issue with website when using Boostrap 5 with **bslib**
  (thanks, @guasi, #2379, @cpsievert, #2384).

- The `tufte_handout()` function inside **rmarkdown** is defunct
  now. Its codebase was moved to the **tufte** package in 2016, and
  this function was marked as deprecated in 2021. Please use
  `tufte::tufte_handout()` instead of
  `rmarkdown::tufte_handout()`. The latter will be removed eventually
  from this package.

- `github_document()` gains `math_method = "default"` and defaults to
  it. No special processing will be done to inline maths in `$` and
  block maths in `$$` as now [Github supports
  it](https://github.blog/changelog/2022-05-19-render-mathematical-expressions-in-markdown/)
  and will render using Mathjax (thanks, @kylebutts, #2361).

rmarkdown 2.14
================================================================================

- Improved highlighting theme `arrow` regarding accessibility.

- Fixed an issue with `site_generator()` detection of a site project
  (thanks, @bhattmaulik, #2344).

- Fixed an issue with Shiny prerendered documents and Pandoc not
  correctly rendering last Markdown paragraph in HTML (thanks,
  @gadenbuie, #2336).

- Fixed a bug that `site_generator()` fails to detect the root dir of
  the site and causes infinite recursion (thanks, @fisher-j, #2339).

- `html_vignette()` gains `code_folding` argument (thanks, @atusy, #2340).

- `html_document()` can opt-out `code_folding = "show"` or `"hide"`
  for individual code blocks by adding the `fold-none` class to the
  code blocks (thanks, @atusy, #2348).

- When using a development version of Pandoc, a `.9999` suffix is
  appended to version number so that `pandoc_available()` can
  correctly compared version with last release.

rmarkdown 2.13
================================================================================
- Fix an issue with older R version and vignette building (#2324).

- Fix an issue with older R version and `preserve_yaml = TRUE` in
  `md_document()` (#2325).

- Long title in `ioslides_presentation` failed to work with Pandoc
  2.17.x (thanks, @Am386DX-40, #2327).

rmarkdown 2.12
================================================================================
- `html_document()` and `html_document_base()` gains the `math_method`
argument to support [all the math rendering engines from
Pandoc] (https://pandoc.org/MANUAL.html #math rendering-in-html):
"mathjax", "katex", "mathml", "webtex", and "gladtex". For backward
compatibility, the `mathjax` argument still works and will take
precedence over `math_method`, but we recommend using the new
`math_method` argument instead of the `mathjax` argument, and the
latter could be deprecated in the future.

  You can specify a math engine via `math_method` as an engine name, e.g.,

  ```yaml
  output:
    html_document:
      math_method: katex
  ```

  or provide both a name and a URL (for `mathjax`, `katex` and `webtex`):

  ```yaml
  output:
    html_document:
      math_method:
        engine: mathjax
        url: https://cdn.jsdelivr.net/npm/mathjax@3/es5/tex-chtml-full.js
  ```

  For `math_method = "katex"`, KaTeX CDN will be inserted in version
  0.15.2 by default (from jsdelivr). A custom URL toward another CDN
  can be passed as `url`.

  For `math_method = "webtex"`, it will default to inset SVG using
  `https://latex.codecogs.com/svg.image?`. Use
  `https://latex.codecogs.com/png.image?` for PNG. See
  https://latex.codecogs.com for supported options (dpi, background,
  ...).

  Most HTML output format using `html_document()` or
  `html_document_base()` as based format should benefit from this new
  feature.  See `?rmarkdown::html_document()` for details (thanks,
  @atusy, #1940).

- `github_document()` also gains the `math_method` argument set to
  `"webtex"` by default so that LaTeX equations can be rendered in the
  Github Markdown document as images. Previously, LaTeX equations were
  not rendered. Set `math_method = NULL` to deactivate.

- Added support for [**katex**](https://docs.ropensci.org/katex/) R
  package as a math engine with `math_method = "r-katex"` in HTML
  documents. This method offers server-side rendering of all the
  equations, which means no JS processing is needed in the browser as
  with usual KaTeX or MathJaX methods. (thanks, @jeroen, #2304).

- `anchor_sections` can now be easily customized using `style` or
  `depth` element for `anchor_sections`. For example:

  ```yaml
  output:
    html_document:
      anchor_sections:
        style: symbol # use symbol style ("hash", "symbol", "icon")
        depth: 2 # max depth to apply anchor on (default to max which is 6)
  ```

  Customizing using a CSS rule is still possible. Detailed explanation
  and examples have been added to the help page `?html_document`.

- Improved the highlighting mechanism in formats that supports `highlight` argument:

  * It is now possible to pass a custom theme file `.theme` in
    `highlight` argument for customizing the [syntax highlighting
    style used by
    Pandoc](https://pandoc.org/MANUAL.html#syntax-highlighting).

  * In addition to Pandoc's own supported themes, two more themes are
    bundled in the package: `highlight: arrow` a theme [optimized for
    accessibility and color contrast](https://www.a11yproject.com/)
    (thanks to @apreshill), and `highlight: rstudio` to mimic the
    RStudio editor theme.

  * For HTML output only, added optional
    [downlit](https://downlit.r-lib.org/) support in `html_document()`
    for R syntax highlighting and autolinking. Use `highlight_downlit
    = TRUE` to activate it (same argument as in **distill**). This
    features require the **downlit** package.

- Templates for `html_document()` and `ioslides_presentation()` gained
  a new CSS rule to display single line `<summary>` content inline
  (rstudio/rstudio#10589).

- `md_document()` gained a new `standalone` argument, which is `FALSE`
  by default unless `toc = TRUE`. This allows to output authors, date
  and other metadata per the Pandoc's template. Due to limitation in
  how Pandoc is handling metadata blocks in its extensions
  `yaml_metadata_block`, `preserve_yaml = TRUE` now deactivate any
  extension to let **rmarkdown** directly handle the keeping of YAML
  block - this means it does not set `standalone = TRUE` by
  default. Meanwhile, `github_document()` gained the `preserve_yaml`
  argument (thanks, @florisvdh, #2297).

- Added `available_templates()` to list all the templates from a
  specific package that can be used with `rmarkdown::draft()`.

- Following support in Pandoc 2.15, `powerpoint_presentation()` gained
  a `incremental` argument as other slide formats. As a reminder,
  setting `incremental = TRUE` will make lists to display
  incrementally. See more in [Pandoc's
  MANUAL](https://pandoc.org/MANUAL.html#incremental-lists).

- Added support for Pandoc's `dir` variable in HTML templates. This is
  the second [Language
  Variables](https://pandoc.org/MANUAL.html#language-variables) after
  `lang`.

- Added a global option `rmarkdown.html_dependency.header_attr`
  (`TRUE` by default). It can be set to `FALSE` to opt-out the HTML
  dependency `html_dependency_header_attrs()` in documents based on
  `html_document_base()` (thanks, @salim-b rstudio/bookdown#865,
  @maelle r-lib/downlit#1538).

- Rendering using `runtime: shiny_prerendered` or `runtime: shinyrmd`
  now natively supports custom templates. Previously since 2.8,
  developers had to add a special comment, `<!-- HEAD_CONTENT -->`,
  conditionally to `shiny-prerendered` variable. (See also NEWS from
  2.8 for the previous behavior). The new behavior inserts required
  special comment `<!-- HEAD_CONTENT -->` as a last element of
  `$header-includes$`. If templates rely on the old behavior and
  require some contents between `$header-includes$` and `<!--
  HEAD_CONTENT -->`, consider including them with `$header-includes$`
  (thanks, @atusy, @gadenbuie #2249).

- A shiny prerendered document with only a empty server context does
  not error anymore. Document will be rendered with a empty server
  function and `server.R` file will be ignored. To use `server.R`, no
  server context should be present in the Rmd document (thanks,
  @jcheng5, #2305).

- Fixed a regression with rendering `shiny_prerendered` document
  (thanks, @aronatkins, @gadenbuie, #2218).

- Fixed an issue in `beamer_presentation()` where `header-includes`
  would be overwritten by `includes = list(in_header =)` (thanks,
  @samcarter, #2294). Same fix as for `pdf_document()` (#1359).

- Fixed broken links to section headers when `number_sections = TRUE`
  is specified in `md_document` and `github_document` (thanks, @atusy,
  #2093).

- `draft()` now works with `devtools::load_all()` and **testthat**
  when used in other packages.

- Lua Filters: Added two more functions in `shared.lua` for other package to use:

  * Added `type()` function backward compatible following Pandoc 2.17 changes.
  * Added `print_debug()` for easier logging during debug.

- Add opt-in `tabset` option to the `html_vignette`. In contrast to
  the similar feature from `html_document`, this feature supports
  navigation from ToC and URL. Any custom formats can introduce this
  feature by using an exported function `html_dependency_tabset()`
  (thanks, @atusy, #2313).

rmarkdown 2.11
================================================================================
- Relative paths in parent directories in the `css` argument of
  `html_document()` were incorrectly normalized to absolute paths by
  #2095 in v2.8. Now relative paths in parent directories will no
  longer be converted to absolute paths (thanks, @daijiang,
  yihui/xaringan#331).

- It is possible to specify the version of jQuery via a global option
  now, e.g., `options(rmarkdown.jquery.version = 2)` (note that the
  default major version is `3`). This is mainly for advanced users and
  developers to test different versions of jQuery.

- `pandoc_citeproc_convert()` now handles correctly bib file
  containing specific UTF-8 characters on non default UTF-8 systems
  like Windows (thanks, @mitchelloharawild, #2195).

- Shiny prerendered documents are now pre-rendered in a child
  environment to avoid allowing the results of static code chunks to
  exist in the Shiny app environment (@gadenbuie, #2203).

- The previously unexported function `convert_ipynb()` is exported now
  (thanks, @acircleda).

(mef)

doc: Updated devel/R-renv to 0.16.0

(mef)

(devel/R-renv) Updated 0.14.0 to 0.16.0

# renv 0.16.0

* `renv` now supports installation of packages with remotes of the form
  `<package>=<remote>`. However, the provided package name is ignored
  and is instead parsed from the remote itself. (#1064)

* `renv` now passes along the headers produced by the `renv.download.headers`
  option when bootstrapping itself in the call to `utils::download.file()`.
  (#1084)

* `renv` now reports if `renv::snapshot()` will change or update the
  version of R recorded in the lockfile. (#1069)

* `renv` now supports the `install.packages.check.source` R option, which
  is used to allow R to query source repositories even if the option
  `options(pkgType = "binary")` is set. (#1074)

* `renv` better handles packages containing git submodules when installed
  from GitHub remotes. (#1075)

* `renv` now handles local sources within the current working directory. (#1079)

* The `renv` sandbox is once again enabled by default. On Unix, the sandbox
  is now created by default within the project's `renv/sandbox` library.
  On Windows, the sandbox is created within the R session's `tempdir()`.
  The path to the `renv` sandbox can be customized via the `RENV_PATHS_SANDBOX`
  environment variable if required.

* Fixed an issue where `renv::status()` could report spurious changes when
  comparing packages installed using `pak` in some cases. (#1070)

* `renv::restore()` now also ensures the project activate script is updated
  after a successful restore. (#1066)

* Fixed an issue where `renv` could attempt to install a package from the
  repository archives even when `type = "binary"` was set. (#1046)

* Fixed an issue where package installation could fail when the R session
  is configured to use multiple repositories, some of which do not provide
  any information on available packages for certain binary arms of the
  repository. (#1045)

* `renv` now uses `jsonlite` for reading lockfiles when loaded. This should
  help ensure useful errors are provided for manually-edited lockfiles
  which contain a JSON parsing error. If the `jsonlite` package is not loaded,
  `renv` will fall back to its own internal JSON parser. (#1027)

* Fixed an issue that would cause `renv` to fail to source the user
  `~/.Rprofile` if it attempted to register global calling handlers,
  e.g. as `prompt::set_prompt()` does. (#1036)

* (Linux only) `renv` now resets ACLs on packages copied to the cache, to
  avoid persisting default ACLs that might have been inherited on packages
  installed into a local project library. If desired, this behavior can be
  disabled by setting the `RENV_CACHE_ACLS` environment variable to `FALSE`.
  If you need finer control over ACLs set on packages moved into the cache,
  consider defining a custom callback via the `renv.cache.callback` R option.
  (#1025)

* Fixed an issue where `.gitignore` inclusion rules for sub-directories were
  not parsed correctly by `renv` for dependency discovery. (#403)

* Fixed an issue where `renv` could report spurious diffs within `renv::status()`
  when comparing package records installed from `pak` versus the default R
  package installer. (#998)

* Fixed an issue where `renv::use_python()` could cause the Requirements field
  for packages in the lockfile to be unintentionally dropped. (#974)

* The R option `renv.cache.callback` can now be set, to run a user-defined
  callback after a package has been copied into the cache. This can be useful
  if you'd like to take some action on the cached package's contents after
  the package has been moved into the cache.

* (Unix only) The `RENV_CACHE_MODE` environment variable can now be used to
  change the access permissions of packages copied into the cache. When set,
  after a package is copied into the cache, `renv` will use `chmod -Rf` to try
  and change the permissions of the cache entry to the requested permissions.

* (Unix only) The `RENV_CACHE_USER` environment variable can now be used to
  change the ownership of folders copied into the cache. When set, after a
  package is copied into the cache, `renv` will use `chown -Rf` to try and
  change the ownership of that cache entry to the requested user account.

* Fixed an issue where repositories containing multiple packages in
  sub-directories could fail to install. (#1016)

# renv 0.15.5

* Fixed an issue where indexing of packages in the package cellar could
  be slow. (#1014)

* Fixed an issue where installation of packages from Bioconductor's binary
  Linux package repositories could fail. (#1013)

* `renv::restore()` now supports restoration of packages installed from
  [R-Forge](https://r-forge.r-project.org/). (#671)

* Fixed an issue where `renv::init(bioconductor = TRUE)` would not update
  the version of Bioconductor used if a project setting had already been
  set.

* It is now possible to "update" an existing lockfile by using
  `renv::snapshot(update = TRUE)`. When set, any records existing in the
  old lockfile, but not the new lockfile, will be preserved. (#727)

* Fixed an issue where `renv` could fail to parse Git remotes for users
  whose username contains a number. (#1004)

* `renv` no longer infers a dependency on the "quarto" R package in projects
  containing a `_quarto.yml` file. (#995)

* Fixed an issue where `renv` could fail to download a package from MRAN if
  a compatible source package of the same version was available from the
  active R repositories. (#990)

* `renv` now reports permission errors during download more clearly. (#985)

* Fixed an issue where `RENV_PATHS_LIBRARY_ROOT_ASIS` was not interpreted
  correctly. (#976)

* `renv::modify()` gains the `changes` argument, which can be used to modify
  a project lockfile non-interactively.

* `renv::clean()` now returns the project directory, as documented. (#922)

* Fixed an issue where `renv` could fail to parse embedded YAML chunk options
  in R Markdown documents. (#963)

* `renv` now sets default placeholder names for the `repos` R option, for
  any repository URLs which happen to be unnamed. (#964)

* Fixed an issue where `renv` didn't respect the `PATH` ordering when
  discovering Python installations via `renv_python_discover()`. (#961)

* Fixed an issue where `renv` could fail to install packages containing
  multibyte unicode characters in their DESCRIPTION file. (#956)

* Fixed detection of Rtools 4.2 (#1002)

# renv 0.15.4

* Fixed an issue where package installation could fail when referencing the
  cache via a tilde-aliased path, e.g. `~/.local/share/renv`. (#953)

# renv 0.15.3

* A variety of fixes for R CMD check.

* `renv` gains an experimental function, `renv::autoload()`, to be used as a
  helper for automatically loading a project for R processes launched within a
  sub-directory of that project. See `?renv::autoload` for more details.

* `renv` will now print a warning message when attempting to read a lockfile
  containing merge conflict markers (as from e.g. a git merge). (#945)

* Fixed an issue where `renv::install()` could install into the wrong library
  path on Windows, if the R installation had a site-wide profile that mutated
  the library paths. (#941)

* Fixed an issue where `renv::install()` would fail to find a package within
  the cache when using an abbreviated commit hash for installation. (#943)

* Fixed an issue where `renv`'s automatic snapshot was not run after calls to
  `renv::install()` in some cases. (#939)

* Fixed an issue where `renv` would incorrectly copy a package from the cache,
  if the cached version of the package and the requested version of the package
  had the same package version, even if they were retrieved from different
  sources. (#938)

* The path to an `renv` tarball can now be set via the environment variable
  `RENV_BOOTSTRAP_TARBALL`, to be used to help `renv` bootstrap from local
  sources. This can either be the path to a directory containing `renv`
  source tarballs, or the path to the tarball itself.

* Fixed an issue where the R site library would not be appropriately masked
  for resumed RStudio sessions. (#936)

# renv 0.15.2

* Fixed issues encountered in R CMD check.

# renv 0.15.1

* Fixed an issue where `renv` inadvertently inserted extra newlines into
  a `DESCRIPTION` file when adding `Remotes` fields after install. (#914)

* Fixed an issue where packages installed from a remote sub-directory would
  fail to install when later re-installed from the cache. (#913)

* `renv` now recognizes YAML chunk options of the form `#| key: value` when
  used in R Markdown documents. (#912)

* Fixed an issue where the R site library was visible in `renv` projects with
  the system library sandbox disabled.

* Fixed an issue where `renv` could update the wrong `.gitignore` in R
  processes launched by `callr` (e.g. in `devtools::install`). (#910)

* Fixed an issue where `renv` could fail to read mis-encoded
  DESCRIPTION files. (#908)

* Fixed an issue where `config$cache.symlinks()` would report `NULL` when
  unset. (#906)

# renv 0.15.0

* The development branch for `renv` has moved from master to main.

* `renv` package records in the lockfile now include a `Requirements`
  entry, which gives a list of R packages this package depends on
  in some way. This is primarily for internal use by the `pak`
  package.

* Fixed an issue where packages containing DESCRIPTION files using
  a latin1 encoding would not be read correctly by `renv`.

* Fixed an issue that could cause `renv::dependencies()` to fail
  when a malformed `DESCRIPTION` file was encountered. (#892)

* The path to the project-local `renv` folder can now be configured
  via the `RENV_PATHS_RENV` environment variable. This can be useful
  if you'd prefer to store your project's `renv` resources within
  an alternate location in the project. (#472)

* `renv` now uses an external library by default for R package projects,
  with the library located within `tools::R_user_dir("renv", "cache")`.
  This directory can be configured via the `RENV_PATHS_LIBRARY_ROOT`
  environment variable if desired. See
  `vignette("packages", package = "renv")` for more details. (#384)

* `renv` now uses the repositories as defined within the project lockfile
  (if any) when attempting to bootstrap itself in a project. (#820)

* The `renv` sandbox is now disabled by default -- see #614 for more details.

* `renv` gains the function `repair()`, to be used to attempt to repair
  the project library when links into the global package cache appear to
  be broken. (#378)

* Fixed an issue where the staging library used during install could fail to
  inherit the same directory permissions as the target library itself. (#331)

* Fixed an regression which caused `renv::history()` to fail. (#886)

* `renv` gains experimental support for the [pak](https://pak.r-lib.org/)
  package. Set `RENV_CONFIG_PAK_ENABLED = TRUE` in an appropriate `.Renviron`
  file to enable `pak` integration. When enabled, calls to `renv::install()`
  will use `pak` to download and install packages.

* `renv::init()` gains the `bioconductor` argument, to be used to initialize
  a project with a particular Bioconductor release. You can also use
  `renv::init(bioconductor = TRUE)` to initialize with the latest-available
  release for the version of R being used.

* Project settings can now be overridden by setting an R option of the same
  name. For example, one could force a project to use Bioconductor 3.14 by
  setting `options(renv.settings.bioconductor.version = "3.14")` within
  the project `.Rprofile` (or similar startup R profile).

* The ad-hoc package repository called "local sources" has been renamed to
  the "package cellar". In addition, the path to the cellar is now
  controlled by the `RENV_PATHS_CELLAR` environment variable, rather than
  `RENV_PATHS_LOCAL`. This change was made to reduce confusion between
  "local sources" of packages located somewhere on the filesystem, as
  opposed to packages explicitly placed in this ad-hoc repository.
  `RENV_PATHS_LOCAL` remains supported for backwards compatibility.

* The `RENV_PATHS_CELLAR` environment variable can now be set to multiple
  paths. Use `;` as a separator between paths; for example,
  `RENV_PATHS_LOCAL=/path/to/sources/v1;/path/to/sources/v2`. (#550)

* Packages installed via e.g. `renv::install("./path/to/package")`
  will now retain the relative path to that package within the lockfile.
  (#873)

* Fixed an issue where invalid `config` option values were not properly
  reported. (#773)

* `renv` now supports restoration of packages installed via one of the
  [r-universe](https://r-universe.dev/) repositories.

* `renv` gains the `bioconductor.version` project setting, used to freeze
  the version of Bioconductor used in a particular project. When set, this
  will override any version that might be selected via the `BiocManager`
  package. (#864)

* `renv` now infers that parameterized R Markdown documents have a dependency
  on the `shiny` package. In addition, R code included within the `params:`
  list will be parsed for dependencies. (#859)

* `renv` now ignores hidden directories during dependency discovery by default.
  If you want to force a particular hidden directory to be included for
  discovery, you can use a `.renvignore` file with an explicit inclusion
  criteria; e.g. `!.hidden/`.

* `renv` now supports the `*release` remotes specifier for GitHub repositories,
  for requesting installation of the latest declared release of a package from
  GitHub. (#792)

* `renv` now handles packages stored within the sub-directory of a Git
  repository better. (#793)

* Fixed an issue where `renv::history()` would fail to resolve the correct
  lockfile path if the working directory was changed. (#834)

* Refined dependency discovery within `glue::glue()` expressions.

* `renv` now parses packages referenced via the `base_format` field within
  an R Markdown document's YAML header. (#854)

* Fixed an issue where `renv` would fail to produce the appropriate binary
  repository URL for RSPM repositories built using Red Hat Enterprise Linux.

* Fixed an issue where `renv::snapshot()` could cause the environment name
  and / or path associated with a Python environment to be omitted from the
  lockfile. (#843)

* Fixed an issue where `renv::restore()` could fail to restore packages which
  referred to their source via an explicit path in the `Source` field. (#849)

* `renv` no longer requires explicit user consent when used within Singularity
  containers. (#824, @kiwiroy)

* `renv` now respects the `R_PROFILE_USER` environment variable, in addition
  to the `user.profile` configuration option, when considering whether the
  user `.Rprofile` should be examined for package dependencies. (#821)

* The `renv` auto-loader can now be disabled by setting the environment
  variable `RENV_AUTOLOADER_ENABLED = FALSE`. This can be useful if you'd like
  to explicitly control how a project is loaded, e.g. by calling `renv::load()`
  explicitly.

* `renv::snapshot()` gains the `repos` argument, to be used to force
  the lockfile to be generated with the requested set of R repositories
  encoded within.

* `renv` now validates that the `repos` option, as used by `renv::snapshot()`,
  is a named vector. (#811)

* Fixed an issue where `renv`'s shims, e.g. for `install.packages()`, failed
  to pass along other optional arguments to the shimmed function correctly.
  (#808)

(mef)