Submitted state.

### Bookkeeper workflow

Bookkeepers can log into the system and see all requests.

When bookkeepers review a Submitted report, they can change the report's

state, and include a note explaining why the report was moved to that state

(e.g., moved back to In Progress because a specific receipt was insufficient

documentation).  When they do this, the system sends email to the requestor

letting them know about the change, including the rationale provided by the

bookkeeper.

The bookkeeper can export any request to the books.  The first release of the

software will simply provide an archive that includes all of the request's

supporting documentation, plus a `.ledger` file with entries for each

expense.  However, note that when building this feature in the code and UI,

it should be relatively generic.  Exporting should remain abstract enough

that integration with other accounting systems remains simple and

straightforward.  Note that even the mechanics could be different; for

example, an SQLedger exporter may add entries to the system directly, rather

than providing the bookkeeper with a file download.

## Requirements potentially for first release

These are features that we would like the system to have, and it may make

sense to make them requirements of the first release depending on how it's

built.

* CiviCRM integration: Many NPOs are already using CiviCRM.  CiviCRM

  integration would provide a familiar interface to users, and simplify

  system administration for the organization.  It may be possible to build

  the system as a CiviCRM extension.  If so, we would get this feature for

  "free."

* Usable without JavaScript: For consistent mission advocacy, it's important

  that some organizations not require requestor to use JavaScript.  (e.g., Tor

  browsers typically have JavaScript disabled because it can undermine Tor's

  anonymity guarantees.)  It should be possible to submit payment

  requests without JavaScript.  The interface can be enhanced when JavaScript

  is available.

  Whether or not we do this in the first release probably depends on what

  framework we decide to build on.  If the framework itself requires

  JavaScript out of the box, it may make sense to have the first release go

  with the flow, then work to add JavaScript-free functionality in a later

  release.

## Requirements for later releases

These are features that we would ultimately like the system to have, but we

know aren't necessary for the first version.  It's good to keep them in mind

when architecting, but also to know that they've been considered and aren't

immediately necessary.

* Allow optional questions: With this, question conditions probably need to

be extended to address the case of "other question isn't answered"

* Additional exporters:

  * Export to SQLedger

  * [Certainly many more, feel free to add them here]

* Richer lifecycle management: A leader may need to approve a request before

  it's added to the books, like an employee's manager or a program director

* Various currency improvements:

   * Automatic currency conversion for validation (e.g., validate that an amount

     in an aribtrary currency is within a limit in USD)

   * Validate currency amounts from outside data sources: The main case for

     this is per diem, where many organizations use rates that are determined

     by another party (e.g., US GSA) and updated periodically.

   * Handle totaling the request based on currency (e.g., expenses can are in

     USD, EUR, CHR but the traveler may want payment in INR.)  Unclear what

     interface for this would look like, but real-time data about past

     currency rates might be available via an API somewhere, and we can use

     that to have the requestor give us "preferred currency for payment" so

     all changes happen in real time in the interface (even allowing the

     requestor to be able to decide *while filling out the report*: "ugh,

     these exchange rates to INR are horrible; I'll have them pay my USD

     account instead").

* Data import: Apps like [Tricky Tripper](http://trickytripper.blogspot.de/)

  let users track expenses for a trip as they go.  The system could import

  this data to prepopulate answers to questions about the request and

  expenses in it.  Probably there would be an import API that can map

  different import formats to a common format, and then administrators can

  define how questions in their system can be answered based on imported

  data.