Yeah, trying the code examples now in a Jupyter Notebook, that I want to keep for reference. But, as already reported, the code doesn't work. That's what I've already been doing for hours beforehand. :) So, the struggle continues. It doesn't help that big restructurings have happened over the years so the models may be getting confused.
As far as the AI goes, I've been paying much more attention to it since being assigned to a chatbot project at work. I've tried various models, including running some locally with Ollama. Cool thing is to use it from a coding assistant. Then a model can analyze the whole codebase. Some vscode extensions were pretty good but they are so chatty that they use up my free allowance for all models very quickly. :)) To summarize, this sucks and I have to keep on looking. :) Strange, but expected to find some working examples quickly. Is it really the case that people don't actually do this that often? In that case, maybe Cashier + Sync will be a nice addition to the ecosystem. I'm adding Beancount support (parsing inputs and outputs) with a switch. On Tuesday, 13 May 2025 at 1:25:02 pm UTC+2 Chary Ev2geny wrote: > I found a very powerful feature available in the OpenAI O3 model, which I > just checked is not available by the Gemini 2.5 Pro: > > I can upload to O3 a zip archive with the project source code, the O3 will > then unzip it, and will be able to explore it on my request. So, in this > way it can analyze entire project, not just code snippets. > > This is what helped me to create this PR > <https://github.com/beancount/beanprice/pull/108>for instance. > > > On Monday, May 12, 2025 at 8:02:10 PM UTC+2 [email protected] wrote: > > Gemini 2.5 pro is killing it these days > - much faster than OpenAI > - better than Claude (anecdotally) > - cheaper than both of these, I think I got a year free with my new phone > - quota seems higher too (I rarely run against the limits compared to the > other models) > Check it out, the web page is ugly, but that's standard Google > cluelessness, the tech behind it is second to none > > > > On Mon, May 12, 2025 at 1:53 PM 'Alen Šiljak' via Beancount < > [email protected]> wrote: > > Thanks!! I was trying with Perplexity and it didn't go well. :)) Maybe it > gets confused by the progressing versions but it was giving some strange > function names in the queries and some non-existent namespaces. > > On Monday, 12 May 2025 at 7:43:15 pm UTC+2 [email protected] wrote: > > https://g.co/gemini/share/ea750735a803 > https://g.co/gemini/share/0905b3166339 > Use the force, Luke! > > > On Mon, May 12, 2025 at 1:37 PM 'Alen Šiljak' via Beancount < > [email protected]> wrote: > > Thank you for this information! > > In the meantime I was poking through the developer documentation. I assume > the Scripting section at > https://beancount.github.io/docs/beancount_scripting_plugins.html > would come in handy, as well? It seems more geared towards plugin authors > so I'll try some more with beanquery. > On Monday, 12 May 2025 at 5:44:08 pm UTC+2 Chary Ev2geny wrote: > > Alen, > > you need to use run_query > > from beanquery.query import run_query > > > You can see how it is done here for instance > > > https://colab.research.google.com/drive/17Mbk2JC4KHMg3b1iXoZg-50kEI2bCoQ-?usp=drive_link > > On Monday, May 12, 2025 at 5:16:22 PM UTC+2 Alen Šiljak wrote: > > Hi all! As I managed to port all my data, I am now adapting the tooling I > built around it, before I can fully switch to Beancount. I would highly > appreciate the assistance from the helpful people in this group. > > I will post the questions separately so that they are easier to track. > > The first question is - how to query Beancount from a Python application? > > To provide some context - I am using Cashier for quick overview and > transaction entry on my phone. I use Cashier Server to synchronize the > balances from the book. Cashier Server is a Python application. This used > to forward queries to ledger and return the data back. Now, with Beancount, > I would assume that it can execute those directly via Python and not run > any external processes. > So, if I manage to identify the BQL queries I need to get accounts, > balances, payees, and whatever else, how would I run them from an external > app? > I have Python experience but I have not worked with any Beancount tools > yet. I've looked briefly into what's available and tried bean-query. > > Thank you for any pointers! > > -- > 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 visit > https://groups.google.com/d/msgid/beancount/7d9315f7-96e1-483e-a791-2e8dc00dda0en%40googlegroups.com > > <https://groups.google.com/d/msgid/beancount/7d9315f7-96e1-483e-a791-2e8dc00dda0en%40googlegroups.com?utm_medium=email&utm_source=footer> > . > > -- > 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 visit > https://groups.google.com/d/msgid/beancount/28bec0d6-2626-40ad-8108-36492fd85745n%40googlegroups.com > > <https://groups.google.com/d/msgid/beancount/28bec0d6-2626-40ad-8108-36492fd85745n%40googlegroups.com?utm_medium=email&utm_source=footer> > . > > -- 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 visit https://groups.google.com/d/msgid/beancount/8817ed74-aca6-4e0d-90b1-ffb758b0a6acn%40googlegroups.com.
