Yeah, back when I ran a mining pool I learned that it is not a good idea to hold all the transactions in memory. Typically you'll rub through the transaction history like a file system journal, applying changes to a dbm to store the last value of each account, and trust in the kernel to cache what should be in memory. Keeping it all in memory sounds aweful, and rerunning the block chain from scratch is not a thing we wanted to do often. -- http://josephholsten.com
> On Jul 24, 2014, at 8:47, Johann Klähn <[email protected]> wrote: > > Interesting idea. But wouldn't this whole endeavor require ridiculous amounts > of memory? > >> On Jul 23, 2014 6:08 AM, "Markus Teufelberger" >> <[email protected]> wrote: >> As far as I understand "triple entry accounting" it extends the 2 standard >> entries from "double entry" by a third "transaction ID" entry that is part >> of a larger system which (ideally) is public knowledge and can be >> independently verified. >> >> This poses a few interesting opportunities but also potential issues: >> Bitcoin for example has until now close to 43 million transactions recorded >> [https://blockchain.info/charts/n-transactions-total] (and some of them are >> not easily decoded or meaningful, so it can be a bit of a hassle to account >> for them correctly, but that's besides the point). If there were a way to >> encode this dataset into ledger (e.g. transactions between >> "Bitcoin:1abcdaddress" accounts) and then referencing the >> transactions/addresses in your actual private ledger, this could be of use >> for additional validation of transactions you enter, something like having >> the entire transaction database of your bank so you can finally make >> sense/use of that transaction ID. >> >> Also it might be a good thing to see some edge/extreme cases of ledger usage >> (e.g. calculating all balances of all addresses from Bitcoin genesis until >> now) with real world data... >> >> Are there actually some hard limits on what ledger can handle (Bitcoin >> transactions are already ~20 GB in binary format, if they get a bit expanded >> to fit into the text format of ledger, they might likely grow beyond 100 GB) >> size wise? Even outputs of a balance command would be several dozen MB of >> data. >> >> All in all it at least sounds like an interesting experiment - I however >> first would like to have some comments about feasability (I know that >> reading the data alone would take quite a bit of time, even from SSDs) and >> ideas where this might be going (some look-aside database that includes all >> transactions, similar to the currency conversion stuff, that gets only >> queried if there is a reference in the actual "private" ledger?). >> -- >> >> --- >> You received this message because you are subscribed to the Google Groups >> "Ledger" group. >> To unsubscribe from this group and stop receiving emails from it, send an >> email to [email protected]. >> For more options, visit https://groups.google.com/d/optout. > > -- > > --- > You received this message because you are subscribed to the Google Groups > "Ledger" group. > To unsubscribe from this group and stop receiving emails from it, send an > email to [email protected]. > For more options, visit https://groups.google.com/d/optout. -- --- You received this message because you are subscribed to the Google Groups "Ledger" group. To unsubscribe from this group and stop receiving emails from it, send an email to [email protected]. For more options, visit https://groups.google.com/d/optout.
