start a Django app

[chadwhitacre]

@nobodxbodon @JessaWitzel Picking up from gratipay/inside.gratipay.com#867 (comment), gratipay/inside.gratipay.com#878 (comment), and gratipay/inside.gratipay.com#871 (comment) ...

When @nobodxbodon and I met last week, we proposed writing a Django application to help us visualize and manage our budget and financials. Our thought was to start by porting over our budget, which is currently in a Google Sheet.

I propose that we add the Django app to this repo, and deploy it to Digital Ocean. What I envision is that we can use Ledger CLI under the hood to run the reports, and we can use Django to read and eventually write the data that Ledger processes. Waddya think? :-)

If you're open to the basic idea, then let's discuss requirements and priorities ...

[chadwhitacre]

Bringing over some relevant comments:

---

That brought us to the tech stack question (3). Long story short, @nobodxbodon is really interested in working on our finances ("If not the most critical part, it's definitely one of them"), and we decided to give Django a try for a finances app that would start out by replacing our budget spreadsheet, and could evolve to take over the access dashboard as well. I'll let @nobodxbodon say more when he's ready.

[…]

We finished by diving into the finances repo to look at what's already there. @nobodxbodon: In addition to digesting the README and trying out the code there, see #308 for the top of a deeeeeeeep rabbit hole that @kaguillera and I went down, where we looked at Xero and eventually landed on the Ledger-based solution I started showing you today)—and then check out #22 for the unfinished conversation about converting to GnuCash.

gratipay/inside.gratipay.com#867 (comment)

---

Here's a mock of the cost part, based on current finance spreadsheet. After discussion, it occurs to me that links of related issues in github can address the "why" and "how" to each cost for business handlers clearly. They serve as sources for (especially external) tech guys to check details and comment, as well as index of internal how-to documents. It's not a strict UI mock of the page, and mainly demo the content that needs to be associated.

gratipay/inside.gratipay.com#878 (comment)

---

As a pretotype, how about I start with some mostly static page(s), instead of a Django app, and see if they look right?

gratipay/inside.gratipay.com#878 (comment)

---

@whit537 as I'm looking at the finance spreadsheet again, it seems the top 3 items makes more than half of the cost. I'm interested in the finer breakdown, like the heroku instances involved, etc. These will be necessary for the mock above as well. My guess is there can be savings somewhere, but I won't be sure unless I see what detailed items are burning money and why. Could you point me to some reference issues please?

gratipay/inside.gratipay.com#878 (comment)

[chadwhitacre]

I just discovered http://plaintextaccounting.org/. :-)

[chadwhitacre]

http://www.ledger-cli.org/3.0/doc/ledger3.html#Budgeting

[chadwhitacre]

Here's what I think we should do:

1. Start keeping records, using the existing Ledger-based system we have today in this repo.
2. Write a Django app (also in this repo) to display the balance sheet and income statement, driven by Ledger under the hood. Deploy it.
3. Move our budget from the spreadsheet into a file in this repo.
4. Add a budget report to the Django app.

What say ye, @kaguillera @nobodxbodon @JessaWitzel, et al.?

[nobodxbodon]

+1
Is there some convention how to keep the records up-to-date? Or will we establish some from now? Because keeping records can be done independently with the app.

[chadwhitacre]

@nobodxbodon Here's the process we came up with:

https://github.com/gratipay/finances#workflow

[JessaWitzel]

How would ledger and this Django app help you reconcile your accounts and make sure that planned spending == actual spending? I use YNAB for my personal budget and find that part to be important. Or will you have to take the extra step to match them up manually?

[JessaWitzel]

Maybe that's not important. Read the docs and per usual I understand enough about it to feel like I don't understand it at all.

[nobodxbodon]

AFAIK there's no planning of spending yet, and the first step would be to book-keep actual spending.

[chadwhitacre]

I'm working on putting the expense report on gratipay/inside.gratipay.com#867 into ledger format for 2016-11 so we can see what this looks like with a practical example.

[chadwhitacre]

PR for that in #30.

[mattbk]

How would ledger and this Django app help you reconcile your accounts and make sure that planned spending == actual spending? I use YNAB for my personal budget and find that part to be important. Or will you have to take the extra step to match them up manually?

I think any type of comparison would be outside of Ledger because that isn't what it's designed to do. How does YNAB do it?

[mattbk]

Does https://github.com/dulaccc/django-accounting/ fit at all?

[chadwhitacre]

Interesting! Hmm ... https://twitter.com/whit537/status/801551260201734146

[kaguillera]

!m @mattbk I was actually looking at that and going to test it out today.
I think that I will look at it on Friday.

[chadwhitacre]

I think an MVP here would be three simple pages:

1. Homepage, with two links to:
2. Income Statement
3. Balance Sheet

(2) and (3) would shell out to the relevant scripts in bin to generate the relevant report, and display it in the browser. To start with this could be just in a pre tag.

Does that sound good to you, @nobodxbodon? :-)

[chadwhitacre]

How would ledger and this Django app help you reconcile your accounts and make sure that planned spending == actual spending? I use YNAB for my personal budget and find that part to be important. Or will you have to take the extra step to match them up manually?

I think any type of comparison would be outside of Ledger because that isn't what it's designed to do. How does YNAB do it?

We can use Ledger for budgeting: #12. Once we had a budget report set up in Ledger we'd then add a third link to the Django app.

[nobodxbodon]

I think an MVP here would be three simple pages:

1. Homepage, with two links to:
2. Income Statement
3. Balance Sheet

Shall we create some charts like mock below (please correct me if I read the report wrong), to make the report easier to understand?

Charts for all the reports may fit into one dashboard all together.

[chadwhitacre]

@nobodxbodon I know for myself I want to see a standard income statement and balance sheet. If we want other reports and charts that's okay too, but, I think we should start with the standard reports. The main point on this PR is to get something together in terms of a Django app that we can deploy over in #29. Release early and often! MVP and iterate! :-)

[nobodxbodon]

we should start with the standard reports

@whit537 +1. I put the mockup mainly to get some feedback.

[chadwhitacre]

The main point on this PR

Erm, sorry. This isn't a PR! :-)

[nobodxbodon]

Trying to create a skeleton django app with a page of balance sheet, but not sure how to get the ledger result in the view. Seems not a good idea to call ledger every time.

[chadwhitacre]

Seems not a good idea to call ledger every time.

I think calling ledger for every request is fine, since a) this is a low-traffic app, and b) Ledger is written in C.

[nobodxbodon]

Now that I actually tried to work out the demo, I'm more inclined to go with existing solutions like fava. There are other options here.

There's already nice tutorial like this to setup a complete personal finanical system (web UI). It can totally apply to here IMO.

[chadwhitacre]

@nobodxbodon Nice find! Fava looks great at first blush—I like seeing the income statement and balance sheet right off the bat in the navigation. What would you see as next steps?

[nobodxbodon]

@whit537 I'll try to setup a demo based on the tutorial like above, on pythonanywhere, with the existing data. In my impression the file format of beancount is very similar to ledger's.

If it works out, we may close this ticket and start setup formally, with the considerations of continuous financial data feeds, etc.

[nobodxbodon]

Just started fava locally with example beancount file from the tutorial (screen below). Appears there's difference between ledger and beancount. Could you go through it to see if it actually suits us?

[chadwhitacre]

Alright, I just read the comparison document, and I am excited to give Beancount a try! I did not appreciate the extent to which it is a mature alternative to Ledger.

[chadwhitacre]

I especially like that it's written in Python 3, and has extensive documentation. The docs give the impression that it has a much tighter design than Ledger:

If people don't want to use them [virtual accounts], that's fine. But Ledger is not an accounting tool; it's a tool that may be used to do accounting. As such, I believe virtual accounts serve a role that others with non-accounting problems may wish to fill.

I respectfully beg to differ.

And even the built-in web interface would be wonderful for our purposes right now:

[chadwhitacre]

Closing in favor of #35.

!m @nobodxbodon