TODO

To-do list
==========


Roughly by order of importance:

- [ ] Build PDF vignettes (see below).
- [ ] Ideally: Compatibility with `dplyr` (see below).
- [ ] Compatibility with `|>`
- [ ] Check interest of `sftime`, especially for intervals:
  https://r-spatial.github.io/sftime/articles/sftime.html#user-defined-time-columns
- [ ] Use `sfheaders`?
- [ ] Check compatibility with `mapview`:
  https://github.com/mablab/sftrack/issues/39
- [ ] Elements of an 's_group' should be atomic:
  https://github.com/mablab/sftrack/issues/32
- [ ] Error model: https://github.com/mablab/sftrack/issues/3
- [ ] Differentiate between point-based and segment-based attributes:
  https://github.com/mablab/sftrack/issues/6
- [ ] 3D coordinates? https://github.com/mablab/sftrack/issues/14


# PDF vignettes

I don't see an automated/easy way to have two versions of the same vignette
alongside (i.e. HTML + PDF). Alternatively, we could build PDF vignettes easily
from the RMD source, e.g. using:

rmarkdown::render("vignettes/my_vignette.Rmd", output_format = "pdf_document",
    output_dir = "inst/doc/", output_options = list(toc = "true"))

That does not require changing anything in the YAML of the vignette source (for
instance, no need to add 'pdf_document' in the list of outputs). I chose
'inst/doc/' as output directory, as all files from 'inst' are simply copied to
the root of the package upon install. That would move the PDF into '/doc/',
where all HTML vignettes built automatically are actually copied when the
package is installed.

The main benefit of this approach is that there is an up-to-date version of the
vignette available at any time, that can easily be added to version control and
provided over GitHub for instance.

I did not test it all the way to install, but that should do it. Now, that means
running the command above manually for every vignette, every time there is an
update of the package. As a solution, I'm pretty sure this could be automated
with GitHub Actions — although I need to get familiar with these.


# Compatibility with `dplyr`

Check:
* https://github.com/r-spatial/sf/blob/main/R/tidyverse.R
* https://github.com/r-spatial/sf/blob/main/R/tidyverse-vctrs.R

See also:
* https://github.com/mablab/sftrack/issues/13