Lobsters: Ledger (2023)

I use a similar tool, hledger(which is a rewrite of ledger in haskell, with more quality of life features). It’s the only tool that stuck with me, and I love it. Plain text accounting is the only tool(methodology?) that allows me to quickly input and edit data because it’s just plain text, and I can use my beloved neovim or tools like grep/sed since it’s all just plain text.

It’s quite handy if you want to know exactly how you spend your money.

P.S: The community is lovely. If you have questions, “what is the way to do it?”, you will get plantly of help if you ask on reddit/forum

I use a similar tool/language, Beancount. Coming from Mint/YNAB/GNUCash, it has been so much easier to wrap my mind around and keep things organised.

With plain text accounting, I now have control of the “mutations” that I make to my spending, vs things being much more opaque with traditional accounting software. This is a super oversimplified analogy, but I like to think of it the difference between functional and imperative programming.

I have been using ledger since 2016, tracking almost every penny, as a hobby. I input data manually or sometimes automatically, e.g. parsing PDFs. The various reports feels like profiling a program, you get insights on your finances. Sometimes I wonder when was the last time I went to a place or bought something, what frequency I go to the hairdresser. It’s all here.

Plain text accounting is very flexible, I often “refactor” / refine my past transactions, with confidence : Emacs + Git + ledger bal. Over time I learned to tune my transactions so that month on month comparisons works well (quarterly or yearly as well). Without these adjustments my history is not as readable/insightful, e.g. an error on my paycheck one month and a resolution the next one could make some discrepancy in my monthly (Income - Expenses) visualization.

Since 2021 I also automatically convert my ledger file to Beancount format (via ledger2beancount) to use the Fava web UI (https://github.com/beancount/fava). An UI with interactive charts is really great to get insights. I filter out exceptional revenues, compare expenses by periods, etc.

Beancount model is a bit nicer/richer than ledger I think, with standardized document directories, ability to link transactions (e.g. doctor expense and reimbursement from health insurance), split of Payee / Narration. I did not commit to Beancount yet because the implementation seems less stable to me. Mainly one author and big rewrites with technologies I’m not necessarily fond of for the use case but symptomatic of a Googler (e.g. protobuf). I hope that does not sound too harsh, Beancount is really great, I just think the implementation is not so “KISS”. I feel like a Rust/Zig/Go implementation, with not too much dependencies, would be easier to hack/rely on.

What I find missing with either tools, is an LSP server, I think there is potentials here, for example when a transaction has an unfilled amount, I’d like for the LSP to show it computed and allow me to fill it in or show the balance of accounts as inlay hints for example.

I started using Ledger on January 1st this year. I thought it would be a chore to keep up, but in the end if I sit down every 2-3 days, not a lot of stuff accumulates, and I don’t spend so much time typing out stuff.

I do very basic reporting- the queries in the linked article are much more than I usually do, ledger bal is more or less useful.

Now I face a quandary- my account structure and tagging are way less than ideal, I think. I wonder if I should start a new structure next year (when I will be more familiar with things), because refactoring my current Ledger file is 2k lines and I dread manipulating it.

If you wonder where you are spending money, it’s quite handy. Now at least I know how much I spend in the supermarket, how much did holidays cost, etc. By showing me that electricity was costing me about 2x more than the other most expensive utility, it pushed me to switch providers, for instance.

It’s pretty nice.

just recently I found the way to prepare for my business tax declaration with hledger by streamlining the tax form fields and (h)leder account names. Really love plaintext accounting. Am curious how it turns out.

Just did this too! Maybe someday the taxman will provide an api for submitting the return so I can automate it.

about the most complicated entries are those split 50/50 private/business and separate VAT, e.g.:

2025-01-13 Easybell.de
    Aktiva:GLS:Giro                   -36,19 €
    e24:eür:43:Easybell.de             15,21 €
    e24:eür:57-vst:Easybell.de          2,89 €
    Ausgaben:Telephon

I calculate them from the csv bank reports with awk into a draft journal and quarterly copy/edit them manually into the journal.

P.S.: e24: is for ‘Elster.de 2024` as I don’t know the form field numbers for the 2025 declaration before end of 2025. So I use the old ones for now and migrate later.

Note: No relation to the cr*pto product with the same name.

Thank god for this immediate announcement.

I also put it in the description of the story, as I had a feeling a lot of people would immediately think of that company instead. :)

I’ve been using ledger for almost a decade now. It’s a great piece of software. I have every transaction in one huge version controlled file.

I used ledger to track ‘virtual’ sub account balances for our joint savings (eg Car, Travel and Renovations)

I’m building a command-line tool to populate the prices.db file (the file that keeps the commodities and exchange prices). For currency exchange rates, I’m using the free tier of fixer.io, and for stock data, I’m using marketstack. The issue with marketstack is that it provides delayed data for the Athens Stock Exchange.

Does anyone know of a service that offers a free-tier API? Even data delayed by 24 hours would work for me, since I only fetch once per day.

I really liked that one too. Usually those little snarky things aren’t super-funny but this was an exceptionally good one.