## Overview

For more details on how to use task views and how to contribute to them, see the repository of the CRAN Task Views Initiative.

## Format

The file format for CRAN task views leverages the R/Markdown format (see Xie, Allaire, Grolemund 2019) so that standard Markdown can be used for formatting and structuring the text and a handful of special R functions are provided to link to CRAN packages, other task views, GitHub projects, etc.

In ctv versions prior to 0.9-0 (released in December 2021), task views used an XML-based format (introduced in Zeileis 2005, R News, 5(1), 39-40 [PDF]). See below for converting task views from the old XML-based format to the new R/Markdown format.

The format is mostly self-explanatory and is illustrated below using an excerpt from the Econometrics task view which is hosted on CRAN at https://CRAN.R-project.org/view=Econometrics and maintained on GitHub at https://github.com/cran-task-views/Econometrics/.

---
name: Econometrics
topic: Econometrics
maintainer: Achim Zeileis, Grant McDermott, Kevin Tappe
email: Achim.Zeileis@R-project.org
version: 2022-09-13
---

Base R ships with a lot of functionality useful for (computational) econometrics,
in particular in the stats package. This functionality is complemented by many
packages on CRAN, a brief overview is given below. There is also a certain
overlap between the tools for econometrics in this view and those in the task
views on r view("Finance"), r view("TimeSeries"), and
r view("CausalInference").

Further information can be formatted with standard Markdown syntax, e.g., for
_emphasizing text_ or showing something really important in **bold face**.
R/Markdown syntax with special functions can be used to link to a standard
package like r pkg("mlogit") or an important "core" package like
r pkg("AER", priority = "core").

- Articles: [Special Volume on "Econometrics in R" in JSS (2008)](http://www.jstatsoft.org/v27/)
- [The Title of a Relevant Homepage](http://path/to/homepage/)

The document structure consists of three main blocks: (1) Some metainformation is given in the YAML header at the beginning (separated by lines with ---), followed by (2) the information in the main text, and (3) a concluding special section called ### Links.

## Details

### Metainformation

The metainformation needs to provide the following elements:

• name gives the name of the task view in CamelCase. This is used as the identifier for installing the view and as the name for the Markdown file, e.g., Econometrics.md, and the auto-generated HTML file, e.g., Econometrics.html. Hence, it should be not too long (typically 1-3 words) and contain no special characters like spaces, hyphens, etc.

• topic is a plain text specification of the topic of the task view. In the example above it is identical to the name but is often a somewhat longer and more detailed title (in title case).

• maintainer gives the name(s) of the maintainer(s) in a comma-separated list. The principal contact should be listed first, followed by a couple of further co-maintainers that help keeping the task view up to date.

• email is the e-mail address of the principal contact or possibly a dedicated mailing list shared by the co-maintainers.

• version is specified by a date in ISO 8601 format (yyyy-mm-dd).

• Additionally, there are may be optional elements: source can be used to link to the source repository (typically on GitHub) and url for the URL of the published task view, respectively. The latter is inserted automatically for the official task views on CRAN.

### Main text

The information in the main text should be a short description of the packages, explaining which packages are useful for which tasks. Standard Markdown format can be used to structure the document with sections, itemized and enumerated lists, bold face, italics, etc.

Additionally, short R code chunks with special functions are used for linking to resources in the same repository: pkg() for normal packages, pkg(..., priority = "core") for important “core” packages, and view() for related task views. A convenience function doi() creates links for DOIs (digital object identifiers).

• The distinction between “normal” and “core” packages is only important for the installation of CRAN task views because the user can specify whether all packages (default) or only the most important core packages should be installed (with all their dependencies).

• If a core package is mentioned several times in the document, it is sufficient to indicate priority = "core" for one of the occurrences.

• Rather than linking to another task view as a whole, e.g., view("Econometrics"), it is also possible to link to specific sections of that view, e.g., view("Econometrics", "Instrumental variables").

Moreover, code projects in other repositories can be linked by using the functions:

Note however that CRAN task views are intended mainly for packages on CRAN (as the name conveys). Thus, links to other repositories should be used for important packages/projects but not list all potentially relevant repositories. Also, it is not necessary to list the GitHub projects for all listed CRAN packages as these are typically provided on the package’s CRAN web page.

## Working with task view files in R

To check whether a task view file has been formatted properly it can be read into R and printed. This should display the metainformation and the list of packages. Subsequently, it can be rendered to an HTML page and displayed in a browser for checking whether the information text is processed correctly. Finally, the function check_ctv_packages() can be used to check whether some of the listed packages are actually not available on CRAN or not currently maintained (archived).

For illustration, the code below employs the Econometrics.md file shipped within the ctv package. Instead a local MyNewView.md with resulting MyNewView.html could be used as well.

library("ctv")
file.copy(system.file("ctv", "Econometrics.md", package = "ctv"), "Econometrics.md")
ctv2html(x)
browseURL("Econometrics.html")
check_ctv_packages("Econometrics.md")

Note that the code above is intended for authors of CRAN task views. For end-users the functions ctv(), available.views(), install.views(), and update.views() are relevant. See https://github.com/cran-task-views/ctv/ for more details.

## Dealing with archived packages

The CRAN packages listed in task views should ideally be maintained actively, so that improved versions are released by the corresponding maintainers in case the daily CRAN checks discover any issues.

However, as even actively maintained packages may be temporarily archived on CRAN, the following strategy is adopted:

• When a CRAN package from a task view is archived, it is still listed in the task view like before. It is only flagged as archived in the text and not installed automatically anymore by install.views() and update.views().

• If the package is still archived after (more than) 60 days, CRAN creates an issue in the GitHub repository of the task view (as in this example).

• At this point the task view maintainers can decide to

• exclude the package from the task view immediately, e.g., if they feel it had not been that relevant and/or not been updated in a very long time;
• reach out to the package maintainer to help with releasing an improved version; or
• wait some more for an improved version, e.g., when they see that the package maintainers already started addressing the problem.
• If the package is still archived after (more than) 100 days, CRAN follows up on the issue and requests removal of the package from the task view. (For sufficiently relevant packages it may be sensible to replace the pkg() link by a github() link in the task view.)

## Converting the legacy XML format to R/Markdown

The ctv package provides an (unexported) R function to facilitate the transition from the legacy XML format for the task view files to the new R/Markdown format described above. Simply using

ctv:::ctv_xml_to_rmd("MyView.ctv")

will create a MyView.md file from the .ctv file. This should do most necessary transformations automatically. However, it is recommended to thoroughly check the resulting file and improve it as appropriate. To check the resulting HTML output use

ctv2html("MyView.md")
browseURL("MyView.html")

Note that one important difference between the XML and R/Markdown format is that the “package list” does not need to be listed separately anymore, it is auto-generated from the “info” text. Similarly, the “links” just need to provide those links that are not auto-generated from the “info” text.