It's not an issue when you are just doing read operations, but one of my normal workflows is a little different. I have a web app with functionality for quickly reviewing, approving, and updating categorization (expense accounts), narrations, and adding additional meta tags (for vacations or other projects). So each time I modify a transaction and update the underlying .bean file, the data needs to be re-read but the .picklecache file is stale so it's a full parse each time. So if I'm reviewing ~150 transactions each month, I might make 50+ edits, each one causing a 3-second delay (I have 50k+ transactions). It quickly gets annoying, but since I do it once a month, it's not terrible.
For the record, I have no interest in switching to a non-Python implementation of beancount just to get improved performance. The benefits of it being Python native far outweigh the parsing speed (for me). On Thursday, March 19, 2026 at 2:19:31 PM UTC-4 Chary Ev2geny wrote: > I actually have the same question. > > I have 20+ years worth of detailed expenses in beancount. Loading ledger > without cash (on updated ledger) takes about 6 sec, with cash 1 sec. > > And, as you said, once it is loaded (in beanquery, Fava or in a notebook > ), all following analysis is quite fast. > > So, taking into account the Moor's law, I have decided, that beancount > speed is not going to be a problem for me personally for the rest of my > life. > > I definitely would not want it to be traded for breaking the API > compatibility > > On Thursday, March 19, 2026 at 5:56:07 PM UTC+1 [email protected] > wrote: > >> I have seen a number of people commenting that Beancount is too slow, and >> there's a desire to see a faster implementation. >> >> I'm not really understanding why this is an issue. Can someone explain >> please? >> >> If you load your beanfile and run the plugins and booking algorithm so >> all the context is sitting in the engine while you make a number of >> queries, why does it matter if it takes a couple of seconds to be ready to >> answer queries? >> >> Or are people running each query as a separate Beancount invocation from >> scratch? >> >> Or are some queries very slow in and of themselves? >> >> >> -- 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/74402398-43cd-41e2-a4e5-115df7e5d51fn%40googlegroups.com.
