#ctrl-c | Logs for 2020-10-03

Back
[05:01:19] <timemachine> AHHH! I'm so happy right now!! I finally got this event store and persistance layer to work correctly.
[05:01:35] <timemachine> I've been hacking away on it all week
[05:26:52] <timemachine> In VS Code, on the SQL Query Results window sone one used the "Trash Can" icon for "Save to CSV" ... lol
[12:49:24] -!- felix [felix!~nttp@212.87.203.208] has joined #ctrl-c
[15:19:06] <calamitous> Awesome! What are you building?
[15:20:30] <felix> o/
[15:22:32] <felix> Who's building stuff?
[15:24:19] <aravk> what kind of stuff
[15:24:30] <slipyx> time machines
[15:25:22] <felix> :D
[15:31:02] <felix> And I just completed a game port.
[15:35:26] <felix> https://notimetoplay.org
[16:58:31] <timemachine> very cool felix
[16:59:44] <felix> Thanks!
[17:02:03] <timemachine> calamitous: it is a CQRS / Event Store. All of the API "Commands" that come in are persisted to a MsSQL Events Table and then dispatched as events that other microservices in the network listen for. the most important event listener is a Persistance layer that updates an optimized MongoDB cluster with the changes from the commands.
[17:03:23] <felix> Sounds sophisticated.
[17:04:24] <timemachine> It is for a fully autible, undo/redo capable, distributed accounting log.
[17:05:29] <felix> I see!
[17:05:56] <aravk> oh man timemachine that's complex and interesting
[17:12:01] <timemachine> We are trying to build a sytem where querying account balances and making rapid machine learning predictions can happen without compromising the original ledger. Which is hard.
[17:12:22] <aravk> ML banking?
[17:12:37] <timemachine> aravk: yup.
[17:12:48] <aravk> what are these predictions going to be used for?
[17:13:17] <timemachine> heh, that's above my pay grade.
[17:13:27] <aravk> sure, sure
[17:13:49] <aravk> but it raises some ethics questions, don't you think?
[17:14:05] <timemachine> you can forcast account trends to better leverage capital.
[17:14:14] <aravk> I see
[17:15:12] <timemachine> also helps with risk/fraud assessment.
[17:21:34] <timemachine> aravk: the current system of just taking someone's word on the value of a loan is more unethical, IMO. ML can prevent the sub-prime loan bundling problems that caused the 2008/09 housing crisis. Those bankers did some shady things and there was no system in place to prevent it.
[17:25:11] -!- emerson [emerson!~emerson@emerson.lol] has joined #ctrl-c
[17:26:10] <felix> o/
[17:26:33] <timemachine> felix, that screen show looks awasome. I've been working on a Rogue-Like in my free time. Using it to learn Rust lang.
[17:26:43] <timemachine> s/show/shot/
[17:26:52] <felix> Thank you! And that sounds good.
[17:27:31] <felix> Mine is in JS, using a binding I just made from Ncurses to Duktape.
[17:28:13] <timemachine> felix: that's pretty sweet.
[17:28:37] <felix> And you're very kind.
[17:31:49] <timemachine> My game is targeting OpenGL right now becasue it make the build/debug cycle fast but I've built it for cursive (rust's bindings for libncurses). I need to make the UI less buggy and then I'll put a build on copper for everyone to look at.
[17:32:07] <felix> I can't wait.
[17:32:42] <felix> Mine is going to be more tricky to build and run since we don't have the Duktape library.
[17:53:59] <aravk> timemachine: I don't know enough to even try to argue anything here; If it makes sense to you, then sure
[18:08:56] <timemachine> :) fair enough. I do agree that we should always be concerned with the privacy implications when talking about ML. It's a bit utopic to cast any AI as a means to give indeviduals to become more autonomous
[18:21:08] <felix> Well, see you!
[18:23:01] -!- felix has quit [Client exited]
[20:26:28] <timemachine> mean while, project "Screen Door" is coming off the honeydew list today.
[21:53:51] <timemachine> doing a merge and I see the infamous, "Friday afternoon commit" commit message :( ... there are 27 files touched by this commit...and that's the only clue I have ...
[23:16:01] <calamitous> eek