just a personal blog really
4,496 karma · joined July 1, 2010
https://djharper.dev
just a personal blog really
I just followed the documentation in here https://beancount.github.io/docs/the_double_entry_counting_m... - it gives you the general principles to follow and you can just pick it up from there.
Double entry book keeping isn't that difficult but that's easy to say once you've been doing it a while
I've been doing PTA since around 2018 and there's definitely lessons I've learned along the way along with plenty of mistakes.
I think the main benefit for me is just the system gives you a complete picture of your finances. The commercial services you can pay for just give you a view into a certain slice (e.g. open banking in UK/Europe to see your current account(s)) - I think mint.com did something similar in the US but it never came over here, I don't know if it still exists. Maybe that's enough for most people, but for me I want everything, investments, liabilities, assets etc. None of these commercial offerings have that because it's so complex and niche, e.g. your open banking provider won't tell you how your pension is doing.
It's also just nice to have the provenance of transactions, e.g. if you receive some shares from work, and you sell the shares and the money ends up in your bank account - the incoming transaction will just be the net proceeds but it won't tell you if you paid any tax prior to that - PTA gives you a more of a complete picture that tracks the whole chain of events that led up that transaction into your bank happening. Overkill for most people? Probably.
As we've crossed into the new year I've switched to a similar directory setup as the OP with 1 file per year. Previously I just had one file that was from 2022 which ended up being like 2 million lines of text, which was starting to bog down the emacs plugin.
What I appreciate the most about this approach to personal finances is it just tracks everything. Investments, pensions, RSUs, bank accounts. You could even go as far as accounting for any resource that's modellable, e.g. energy usage in kwh vs. bills. I probably wouldn't go that far though :D
Also you can build a bunch of tooling around it too, with the advent of LLMs my toolset for beancount management has expanded quite significantly. Most recently I got claude to rewrite my transaction rules engine https://djharper.dev/post/2025/08/19/using-llms-to-turn-scri... into something nicer with a UI. This would have taken days to build in the before times, and I probably would not have bothered because it's overkill for 1 user (me)
Windows 11 has been fine for me, I don't interact with it much other than seeing it for a bit when launching games.
I honestly wouldn't mind giving Linux a go, the only downside is I made the mistake of buying an nvidia graphics card, I'm not sure how much of a pain it is these days but last time I tried it was a bit of a nightmare - the general wisdom at the time was to go with an AMD card.
> Workers download, decompress, and materialize their shards into DuckDB databases built from Parquet files.
I'm interested to know whether the 5s query time includes this materialization step of downloading the files etc, or is this result from workers that have been "pre-warmed". Also is the data in DuckDB in memory or on disk?
I don't use any of the dynamic features though like labels in docker containers etc, all of it is configured using the static configuration. It's been working well but I don't think about it really.
Which got me thinking about how do these frontier AI models work when you (as a user) run a query. Does your query just go to one big box with lots of GPUs attached and it runs in a similar way, but much faster? Do these AI companies write about how their infra works?
This has been working great for the occasional use, I'd probably top up my account by $10 every few months. I figured the amount of tokens I use is vastly smaller than the packaged plans so it made sense to go with the cheaper, pay-as-you-go approach.
But since I've started dabbling in tooling like Claude Code, hoo-boy those tokens burn _fast_, like really fast. Yesterday I somehow burned through $5 of tokens in the space of about 15 minutes. I mean, sure, the Code tool is vastly different to asking an LLM about a certain topic, but I wasn't expecting such a huge leap, a lot of the token usage is masked from you I guess wrapped up in the ever increasing context + back/forth tool orchestration, but still
My setup is quite simple, it's just a few VMs with one docker compose file for each. I have an ansible playbook that copies the docker compose files across and that's it. There's really nothing more to it then that, and maintenance is just upgrading the OS (Fedora Server) once the version reaches EOL. I tend to stay 1 version behind the release cycle so upgrade whenever that gets bumped.
I do use nix-darwin on my macs so I do _see_ the value of using a nix configuration, but I find it difficult to see if the effort in porting my setup to Nix is worth it in the long run, configuration files don't get written in a short time. Maybe LLMs could speed this up, but I just don't have it in me right now to make that leap
> No, we are not asking that at all. We are in favor of publishers ending support for a game whenever they choose. What we are asking for is that they implement an end-of-life plan to modify or patch the game so that it can run on customer systems with no further support from the company being necessary. We agree that it is unrealistic to expect companies to support games indefinitely and do not advocate for that in any way.
It does make me wonder whether we'll start to see more and more computers with unified memory architecture (like the Mac) - I know nvidia have the Digits thing which has been renamed to something else
One thing that stood out to me in the article was this this to justify the shutdown
> But the way people use the web has evolved, so we’re channeling our resources into projects that better match their browsing habits and online needs.
I'd be really interested to hear what exactly they mean by this, are people visiting fewer websites? Walled gardens like facebook etc make it useless for bookmarking so I can see how pocket would be a bad fit there
In my experience the algorithmic recommendation systems don't do this, I mean they might throw you a wildcard in here or there but I tend to find they overfit on some niche and it just becomes tiresome, and you don't get the commentary from the DJ who might add something like describing who the artist is, what the song's name is and maybe some flavour on the DJs interactions with the artist over time.
Although, given Amazon’s recent efforts with the kindle, I’m going to say they will probably remove the ability to download books soon
I thought at first people were just ignoring me, but when a company reached out to me over SMS to respond to a complaint I had, they said their email reply had bounced so was contacting me on SMS instead
Switched to fastmail at that point.
In the north though it’s a mixed bag, frequently delayed, huge underinvestment, expensive etc.
The whole cloud computing world was built on hypervisors and CPU virtualization, I wonder if we'll see a similar set of innovations for GPUs at commodity level pricing. Maybe a completely different hardware platform will emerge to replace the GPU for these inference workloads. I remember reading about Google's TPU hardware and was thinking that would be the thing - but I've never seen anyone other than Google talk about it.
At the end I was losing track of my mouse cursor and fatigue was setting in, the final ending scene was such a calming sense of relief!
Quite a bit of a ramp up to get used to it, but no other tool outside of spreadsheets comes close to flexibility.
Do you worry about the upgrade cycle on the hardware? Can't be fun replacing the CPU in lots of machines :D
I believe the V2 models have updated firmware to support IPV6 firewalling but I made the mistake of buying the V1
It all worked ok until I noticed I could access open ports on my Mac from the public internet, and then found out my router does not support blocking incoming traffic for IPV6.
Granted, it’s probably unfeasible for attackers to scan the entire space to find open ports, but I still wasn’t comfortable with leaving all my stuff exposed, so I disabled ipv6 again and came crawling back to v4.
Maybe when/if I get a better router that supports firewalling ipv6 I’ll try again.