English
资源说明：Double-Booked: a Chrome app for double-entry accounting.
# Double-Booked: A reactive web app for double-entry accounting
This is an app for double-entry accounting, in the tradition
of [GnuCash](http://www.gnucash.org/) and
[Ledger](http://www.ledger-cli.org/). It is designed in a
reactive style, with the goal of having a highly responsive
UI where updates propagate immediately and you don't
"generate a report", you just view it (with instant
results).
It is designed to run in a web browser, client side, using
indexedDB as its data store (though with the ability to sync
to a centralized server).
I have lots of big ideas for how I want to improve on
existing accounting apps. But as a backend guy, I am
primarily focused right now on creating a data storage layer
that has all the features I want. I want to make sure the
data model is rich and powerful (hierarchical accounts,
multiple currencies, stocks/bonds with lot/basis info,
balance assertions), but also highly efficient. Propagating
updates efficiently in an accounting ledger is a hard
problem that I discuss at length in
[DBDesign.md](DBDesign.md).
Basically I'm trying to create the "git" of double-entry
accounting. Git's primary contribution is its efficient
data model and storage format. It is the basis on which the
entire ecosystem is built.
# Build / Run / Test
You need [Node.js](https://nodejs.org/).
Install these tools if you don't have them globally
installed already:
$ npm install -g bower
$ npm install -g gulp
Then install our dependencies into the local tree (it reads
the list of packages to install from `bower.json` and
`package.json`):
$ bower install
$ npm install
Now build/transpile all the source and create a
fully-populated output directory in `build/`:
$ gulp
Now you can run the app by visiting `build/index.html`
(not much to see right now though).
To run the tests visit `tests/test.html`. I have verified
they work in both Chrome 43 and Firefox 39.
# Overall Design / Technologies
The data storage layer is in [model](model) and written in
ES6. It only depends on:
* a red-black tree, since JavaScript doesn't natively
provide a sorted map.
* indexedDB, for storing the data locally
Because of the second dependency, it requires a web browser
(there doesn't appear to be a robust implementation of
indexedDB for Node.js).
The storage layer is by far the most interesting and mature
part of the project so far. Between the [design
doc](DBDesign.md) and the source comments, it should be
pretty usable. It is also tested reasonably well by the
tests in `tests`.
The (very minimal) UI is built on
[React](http://facebook.github.io/react/) and
[ReactRouter](https://github.com/rackt/react-router). It
uses some minimal UI framework in the form of
[Pure](http://purecss.io/) and
[FontAwesome](http://fortawesome.github.io/Font-Awesome/).
Since the app is designed to be capable of working
completely offline, it is intended to be usable as a Chrome
App or with ServiceWorker. I'm keeping my options open on
this point.
My hope is to implement syncing with a server using a flat
file on Google Drive or DropBox. So you'd have a local
database in indexedDB, but you could push your commit log to
Google Drive and pull commits from it. A key idea here is
that it should sync to storage owned and controlled by the
user, not servers that I operate.
本源码包内暂不包含可直接显示的源代码文件，请下载源码包。
