---
title: "zoo Design"
output: rmarkdown::html_vignette
vignette: >
  %\VignetteIndexEntry{zoo Design}
  %\VignetteEngine{knitr::rmarkdown}
  %\VignetteDepends{zoo}
  %\VignetteKeywords{irregular time series, ordered observations, time index}
  %\VignettePackage{zoo}
---

This is a set of design principles that - albeit not having been explicitly set out
initially - have guided the development of the R `zoo` package.


## zoo works with any ordered index class

... having the prescribed methods, see [?zoo](https://zeileis.codeberg.page/zoo/man/zoo.html) and
Zeileis and Grothendieck (2005, _Journal of Statistical Software_, [doi:10.18637/jss.v014.i06](https://doi.org/10.18637/jss.v014.i06)).
Specific index classes are not hard-coded in functions (with
a few exceptions necessitated by external interfaces such as reading
and interfacing with `ts`) but rather all index operations are only done
via the allowed methods.


## zoo is invisible

As far as possible new names (e.g., for functions or methods
and their arguments) are not introduced.
New functionality is added as methods to generics from base R
allowing reuse of those names. When names are added, thought is put
into the name since `zoo` is so miserly about adding them.  For example,
the `read.zoo` name is obviously derived from `read.table` and the names
`na.locf` or `na.approx` are in analogy to other `na.*` functions
frome base R.


## zoo is consistent with base R

The idea is to make the usage of `zoo` completely natural to someone who already has experience with R
(in particular with the `ts` class).  Certainly, `zoo` contains extensions
to base R, but they should be as consistent as possible and have a natural
"look and feel".


## zoo is infrastructure

`zoo` tries to do basic things well, e.g.,
data reading, handling, aggregation, transformation, etc. However, it does not
provide time series modeling functionality - `zoo` rather encourages add-on
packages. A list of current packages employing `zoo` can be found on the Comprehensive
R Archive Network (CRAN) and in the `zoo` FAQ, see `vignette("zoo-faq", package = "zoo")`.


## zoo interfaces to all other time series packages

... on CRAN (or at least all of the more popular ones and some of the relatively unknown
ones as well). Thus, time series information can be passed back and forth relatively easily
between `zoo` and other time series classes and hence combinations of various
functionalities are facilitated.


## zoo includes extensive documentation

In addition to the standard `help()`
pages, `zoo` provides a set of vignettes, see `vignette(package = "zoo")` for a list.
Standard help and vignettes are also available on the project web page at
<https://zeileis.codeberg.page/zoo/>.
Furthermore, the `zoo` Development Team members also post frequently
on StackOverflow and the R mailing lists whose archives hence contain many useful code snippets etc.


## zoo long had no bug list since all bugs are fixed (almost) immediately

Over the years, `zoo` has grown into a mature package. Nevertheless, there are always still bugs
(or border cases that are not handled well). When reported to the development team,
these typically get fixed quickly. Reports can be made via e-mail to the maintainer
list or reported in the code repository at Codeberg.org: <https://codeberg.org/zeileis/zoo/issues>.


## zoo includes formal regression testing

... using R's own test system (in `zoo/tests/`).