Want to share your content on R-bloggers? click here if you have a blog, or here if you don't.
Last summer, just prior to starting as an intern working for Hadley, I published
fiery
, a flexible webserver framework
that I had worked on during the spring. The reason why I developed fiery
amids
offerings such as shiny
, opencpu
, and plumbr
was that I was missing a
back-to-basic server framework with low overhead and freedom to program your
server logic in the way you wanted. For all of their accomplishments I felt that
the other frameworks had different goals.
fiery
has always been intended to be modular, with support for adding plugins
in order to augment functionality, but since publishing fiery
I’ve
unfortunately been distracted by other projects (mainly ggforce
, ggraph
,
tidygraph
, and particles
). During all this my bad conciusness about the
fiery
standstill has nagged my though, and for several different reasons I’ve
decided that now is the time to restart development of this ecosystem. This post
is as much a help and incentiviser for me as it is an overview of my plans for
prospective users.
routr
The first piece of the puzzle to work on is my
routr
package which was intented to be
released rather shortly after fiery
itself but fell down the priority list as
my visualisation packages gained traction. While the main functionality of the
package is there, it’s rough around the edges and is in general need of a full
code review and update. Once routr
is completed the whole act of writing
server logic in fiery
will be much more streamlined and easy.
A Deployment scheme for fiery apps
fiery
is both intended for local running apps in the same vien as htmlwidgets
etc., for powering webapp server logic, and for services deployed to production
environment. For the two latter there has yet to come a streamlined deployment
scheme and this is of course a prerequisit for usage. My current plans, which
will be developed under the firestorm
mantle, is to set up infrastructure for
easily deploying fiery
apps as docker images and leveraging as much as
possible from the docker ecosystem in terms of setting up load balancing and
service discovery. Optimally all of this should be doable from within R and
without fiddling with dockerfiles and other configurations.
Making (some) security easy
The internet is a wild place and you shouldn’t expose a sevice without at least
some security measures. While not all of these shld be implemented at the
fiery
layer, those that do should be as easy and obvious as possble. The
firesafety
plugin package will attempt to collect all the best practices for
secury http(s) and websocket communication with the big wide world
Taking caree of your data
Whether you’re building a stateful app where session should be managed, or a
REST api, data needs to go somewhere. The firesale
plugin package will be the
home of a fiery-centric datastore implementation, with support for sessions and
the likes.
Demos for the demo god
No web framework is complete without example apps, hopefully along with easy-to-read guides on how to build them. I hope to be able to provide this as well down the line, though focus is of course on providing good R documentation.
Wrapping up
That was a lot of words and promise without anything to really show for it. I’m acutely aware of that. The above provides a very loose overview of what I intend to do, but I’m in no position to provide a timeframe – you can read the list as sort-of prioritised thus getting a sense of where my spare time will be spend first.
Here’s to hoping I don’t get derailed again ?
R-bloggers.com offers daily e-mail updates about R news and tutorials about learning R and many other topics. Click here if you're looking to post or find an R/data-science job.
Want to share your content on R-bloggers? click here if you have a blog, or here if you don't.