Hello Dustin, On 07/02/2021 16:23, Dustin Farris wrote: > I've been using beancount for a little over a year to manage my personal > finances as well as some side self-employment accounting. I have twice > gotten frustrated to the point of trying other products (specifically > Personal Capital, Mint, Quicken, and QuickBooks) but give up and come > back to command-line accounting for reasons probably familiar to > everyone in this group.
It seems that most of your frustrations are not rooted in how Beancount works, but are > So here are my frustrations and what I'm going to attempt to do for > myself to address them. I'm mainly venting here, but if anyone has any > suggestions I'd really appreciate them. > > * I update my journal every month. Getting updated transactions from > 25+ different accounts every month is very time consuming. This is not specific to Beancount, is it? As it has been mentioned by others there are services that do the data aggregation for you, if trusting an online app with your accounts credentials is something for you. One thing I can suggest is to take an incremental approach to the number of accounts you import, to the frequency at which you do it, and to the amount of detail you record. Having less to manage while you build your workflow will make things easier. It is easy to go back and add information when you need it. > o Todo: Research programmatic downloads of transactions from all > banks (has this been done already?) Back when I started working with Ledger (before switching to Beancount) I wrote some code to download statements for some of my accounts. Then I realized that writing them and keeping them working takes much more time than I would ever save compared to downloading manually once a month or so. Even more so nowt that I have two-factor authentication for all my accounts. > * The beancount file is getting overwhelmingly large after just 1 > year. This is making it hard for me to jump around and find/fix > things. I often have personal transactions that cross equity > accounts into our rental business, or my software engineering self > employment. > o Todo: split personal.beancount into smaller journals (by month?) Also this does not seems to be an issue specific to Beancount. Once you have a lot of data (for some definition of "a lot") you need to have a way to sort through it. I don't know of any system that would solve this magically. You need to identify what is the key you most often would like to use to look-up a transaction and organize your data accordingly and maybe write tools to help (bean-doctor could serve as an example, if you are thinking about writing some tools to find things in your ledger). I use a single gigantic Beancount file where transactions are sorted by date. I have outline-mode headers that split it up by year and month, but those mostly help when I cut-and-paste in the imported transactions. Most often, when I am looking up a transactions, I know the date or the time range in which it occurred. I have an Emacs function that (assuming the ledger is sorted by date) looks up a specific date. I use this both for inserting transactions manually and for looking up transactions. It is rare that I have to manually amend transactions once they are imported > * omni-complete in vim is an awkward keyboard chord and account > completions in vim are sometimes broken depending on what you last typed You should try Emacs and beancount-mode :-) > * I still don’t understand how reporting works, and part of that is > because I don't use it enough. I feel like I'm relearning Beancount > Query Language every time I do need something. > o Todo: keep a list of commonly-used queries As it has been suggested already, you can have queries recorded in the ledger itself with the "query" directive. > * reconciling receipts / splitting transactions is time consuming, > although vim macros help here > o Todo: Improve importer recognition of payees and likely expense > accounts This is also not Beancount specific. I actually thing that the text ledger format helps immensely here (assuming you use a text editor that cooperated and not one that stands in the way). Probably some other application embed some machine learning that partially automates this, but you can have the same for Beancount imports. There is smart_importer, or you can roll your own. I have my own solution leveraging some ML in scikit-learn and it is less than 50 lines of code (it is so short that I never really felt it worth to be published). The rule-based system I was using before is more lines of code. > * no way to attach receipt pictures to transactions? > o Todo: Research beancount tooling or other apps to capture receipts You can associate metadata to transactions. Beancount does not interpret it, but you can put the path to a receipt picture in there. An Emacs function to display the receipt would probably be around 10 lines of emacs-lisp. > * fava is nice, but could be nicer > o Todo: Research writing a new frontend reporting/visualization > tool and/or contribute to fava Every nice thing could be nicer :-) Cheers, Dan -- You received this message because you are subscribed to the Google Groups "Beancount" group. To unsubscribe from this group and stop receiving emails from it, send an email to [email protected]. To view this discussion on the web visit https://groups.google.com/d/msgid/beancount/a059306a-8e29-f990-a1cc-ae031f9626e6%40grinta.net.
