-
Process all sections instead of only the first (#25, #36).
-
Fix
\href{}{}
macro and add\dots
and\ldots
to single-argument macros (#22, #36).
- Conversion fails if function name contains a regex metacharacter (thanks, @joshuaulrich, #32).
- Fixed a
parse_file()
error that can occur when\keywords{}
is empty in.Rd
(thanks, @byandell, #29).
- Reduced one empty line in the examples sections generated by the roxygen tag
@examplesIf
.
- Supports the roxygen tag
@examplesIf
.
- when writing out the Rd file to document re-exported objects, the links are changed from
\link[pkg:name]{name}
to\link[pkg:topic]{name}
, due to the recent requirement of CRAN that links should point to the topic instead of name
roxygen_and_build()
now callsroxygen2::roxygenize()
in a new R session.
rab()
no longer reformats code in the usage sections of Rd files.
- No longer imports formatR in NAMESPACE (using formatR::tidy_source() now).
- importRd() fails due to a change in roxygen2 (r-lib/roxygen2#568).
- the roxygen2 comments in Rd and NAMESPACE are no longer removed
- existing tarballs of the source package will be deleted before building the new tarball with
rab()
-
rab(install = TRUE)
can automatically generate an Rd file for objects that are imported from other packages and later re-exported in the current package; this is done through the internal functionimportRd()
-
when
build = FALSE
andinstall = TRUE
inrab()
, the package directory instead of the built tar ball is installed -
added an argument
install.opts
torab()
to specify options forR CMD INSTALL
(default:--with-keep.source
ifbuild = FALSE
, otherwise empty) -
added an argument
before
torab()
so we can evaluate an R expression before the package is roxygenized and built -
if the Examples section starts with a special token
# !formatR
, the example code will not be reformatted -
spaces will be added after commas in
\item{}
, e.g.\item{x,y,z}{}
will be converted to\item{x, y, z}{}
- make sure the correct R executable is used (i.e. always use file.path(R.home('bin'), 'R'))
- added a new argument 'build.opts' to rab() to pass options to R CMD build
- the new feature @returnItem introduced in Rd2roxygen 1.1 was temporarily removed, since it is not supported in roxygen2 (3.0.0) yet
- Rd2roxygen depends on formatR v0.9 for better handling of example and usage code
-
added NULL to create_roxygen() when docType is data or package (#9)
-
Rd2roxygen() tries to add the @export tags according to the exported names in the NAMESPACE (#10, #11 and #12)
-
when reformatting usage and example sections, rab() tries to make code width smaller than 90, as required by R CMD check
- the package vignette is built with knitr alone now (under R 3.0.x)
-
\item{}
in the\value{}
section will be formatted as the @returnItem tag, which is a new tag in roxygen2 (> 2.2.2) (#8) (thanks, Andreas Alfons) -
R code in
\dontrun{}
,\dontshow{}
and\donttest{}
can be reformatted by formatR now
-
the vignette is built with knitr via the Makefile; in previous versions, it was a fake Rnw document (essentially a weaved tex file)
-
dependency on roxygen2 and formatR was changed to Imports
-
default value for 'check.opts' in rab() is '--as-cran' now
- @section is supported now:
\section{title}{content}
will be converted to##' @section title: content
(thanks, Brian G. Peterson)
- the package vignette now compiles with knitr instead of Sweave
\dontrun{}
can be correctly reformatted now; there was a bug in the previous version when options(replace.assign = TRUE)
- usage for S3 methods like
\method{generic}{class}(param = value)
can be correctly reformatted now; in previous versions rab() will ignore such usage sections
-
Rd files that have
\docType{data}
or\docType{package}
(for datasets and package documentation) are written into xxx-package.R directly; these names are not looked up in R scripts under the R directory of the source package -
Rd comments (following %) will be ignored when generating roxygen comments for arguments
-
a complete example of running Rd2roxygen() on a real package was added to ?Rd2roxygen
-
re-run Rd2roxygen() on a package will remove the roxygen comments in R scripts before adding new roxygen comments parsed from Rd files; in old versions of this package, the old roxygen comments were not removed, hence the comments were added incrementally above the function objects, which was certainly not desirable
- all these forms of quotes in assignments are allowed in Rd2roxygen():
foo
<-, 'foo' <-, "foo" <-, and = can be used as the assigning operator as well
- the usage sections in Rd files will be reformatted by default again when calling rab() (this feature was formerly removed in version 1.0-0)
- rab(..., install = TRUE) will install the *.tar.gz file rather than the package directory; similarly, rab(..., check = TRUE) checks the tar ball as well
- the percent symbols in the examples code will be unescaped (i.e.
\%
becomes%
) before the code is reformatted, then escaped again; this can avoid errors when reformatting the code, since\%
is an illegal character in R
- Rd2roxygen uses the brand-new roxygen2 rather than the old roxygen package which is no longer maintained; users are recommended to read the NEWS file of roxygen2 carefully. Some major changes include: the description is in the second paragraph rather than the second line now (i.e. there is an empty line between the title and description); @format and @source are supported, and a series of new tags are introduced in roxygen2 too
-
the escape and use.Rd2 arguments are removed in rab() since roxygen2 can handle these cases elegantly
-
roxygen.dir is the same as the package directory by default, so rab() will write all the documentations in the original package directory
-
the usage section is no longer reformatted by formatR; roxygen2 formats it nicely
-
the function rm_undocumented() is no longer needed, so it was removed from this package; use @noRd if you do not want the documentation of a certain object to be generated by roxygen2