วีดีโอ

19:29 banger

5:07 "you might even lose some hair if you try to deal with that"

The link provided in the description does not work btw.

I do walk&talk technique in other areas of my life as it helps me with thinking. I LOVE it, but others really HATE it. :D

Chipper(*), competent, dirty hands w/OSS, and doesn't wear her genius on her sleeve. IT needs more peeps like this. (*) - admittedly, somewhat less so once i realized I'd been listening on 1.5x and slowed down, but still....

heresy! boo hsss

He cut his hair ❤❤❤

Why wouldn't it be deterministic? If the algorithm is replicated then it should be. Even if an RNG were used, you can assure that the same random sequence is used.

Really changes the way u think about coding

19:26 - So good (on top of the interesting talk!) - Thanks Andrew for putting this out there!

What if Jesus was a database and comes back tomorrow?

So really - have design phase be important and assign more time to it - assert errors that can happen due to programmer errors and not simply for recoverable errors - build simulation tools that can stress the system and bring things down.

On Unix, you can use lpr/lpd/etc to implement an audio player and queue. 😎

I love to have that wallpaper

TigerLang when?

We used this style in university, it was called defensive programming.

Just starting with TigerStyle. Baby steps though. Trying to add at least two assertion per function.

i didnt understand a lot, but maybe in a year i will

First of all, thank you for this series, really really informative!! I just wanted to ask what is it that you use for searching all files in VSCode?

Learnt so much! Kinda like the silver bullet or the promise of agile/scrum :)

You tout Zig's default assertions for arithmetic overflow, but 5 minutes earlier you dinged languages that turn off assertions in production builds--which Zig does. The inconsistency is due to talking (and thinking) in terms of absolutes. The reality is that there's a tradeoff between performance and safety, and the Zig compiler makes the often reasonable assumption that frequently executed assertions like arithmetic and buffer overflow are too expensive to retain in production. But the language could do a lot better--the length of an array should be constrained to a range type, and indices should be constrained to a range type with min and max 1 less than the length type. Always check that values with a range type are valid. Check that the range type of an index value is within the actual length of the array--such checks can be done once when the array length is known--certainly lifted out of loops. Actual index bounds checks can be eliminated. For arithmetic overflow, we just need control over which blocks are checked under which build types. By default they should be on for debug builds and off for production builds. BTW, after listening to the whole talk, I still don't know what "Tiger Style" is, but did hear some stuff about sculpting and painting that struck me as irrelevant filler.

Conversational transactions == State the problem is micro services sold to masses - that's the bad abstraction

12 minutes in. so far nothing was said.

Amazing talk

Why even do DDL inside transaction. This is should be a BUG

I'll probably have to think about this for the next week. It can surely change the way in which I program

These guys are probably the only devs openly talking about how to do event sourcing and machine sympathy correctly there are others but they keep their success/research to themselves this is gold in the right hands

Louis CK is one of the best programmers I know!

i aspire to be as enthusiastic about my projects as joran is about his

wow! Very good talk!

Have couple of questions; 1. How do you develop new versions? 2. (to be continued)

This makes me think of the RethinkDB query language with explicit index range.

if someone comes here wondering - like I was - about what the animation is showing when talking about quicksort, I believe it's the Lomuto partitioning.

The thousandth like was given

The Grand recap without Tracy?😢

19:51 DATOMIC

Amazing talk! Thank you Kyle for helping us fix this in MariaDB. The good news is that, since version 10.6.18, MariaDB includes the variable `innodb_snapshot_isolation`.

Awesome talk. Not a database guy, but learned a lot about DBs just by watching to Joran's talks

Awesome facts. Thanks, Joran :)

Such a great talk. Thanks Joran!

Great talk! Loved the pragmatic approach to the system and trade-offs of working to enable a product focussed company. Aligns a lot with what I've been experiencing at my work, all the lessons learned rang incredibly true to me.

Insightful talk, I would love to see better alternatives to JWT become common. However server side sessions still seem like a solid alternative, with far fewer problems/attack surface than any of the signed token approaches discussed.

Fastly have done some fantastic work in this area, it's great to hear about it! I would like to warn about putting more than one capability into a credential. Once you do this, the lovely mathematical properties that object-capability systems grant you don't entirely hold, even if the weakness is nicely scoped. Being able to re-use one credential for a different resource is at the root of a wide variety of vulnerabilities, especially attacks involving enumeration or ibac.

Isn’t the fact that he has a black card authorization and not authentication?

yes but can i test a sum function with this thing? lol

You have to look into meta data checksums in ZFS as it enables bitrot correction. Unlike Linux, Solaris and later BSD added ZFS exactly because of oblivious data loss caused by bitrot. And get this, ZFS was introduced with Solaris 10 a full decade before Fsyncgate… While you can’t easily update the IO used in a database, you can install a BSD distribution and format your database storage disk array with ZFS for handling all mission critical data. An enterprise would probably go with Oracle DB on Solaris . That said, for financial data, TigerBeetle is currently the single best option because not everyone has the time and expertise to gronk BSD / ZFS or the money for Oracle on Solaris.

Dang we got a Borges reference... sign me up!

You're Telling Me: Database Scaling - Because You Want Me to Believe its Cheap to Transmit Data on your Networks. I can come To America. build a Website in My Garage. Steal somethings from your Basement and Have a Successful E-Commerce Web Store. So You're Telling Me Database Scaling: "Come explore lessons learned from 5 years of building databases at petabyte scale at Dropbox and Figma" - You Want Me To Come for an Education at Dropbox/Figma. Where You'll Teach a Pr0stitute of Yours to Fu*ck Me.

You're Telling Me: Database Scaling - To Hide or Prevent Me from Learning - How You've Built Your Satellite Broadband Networks.

The Problem isnt Database Scaling. You Simply Shard the Database. Shard Means Split Up. Customers With surnames Beginning With A, are On Database 1. Customers With surnames Beginning With B are On Database 2. you Can Shard at The App Layer: PHP or ASP or You Can shard at The Database Layer: PGPool (Proxy DB). The Difficulty is Building The Network. You Have To Pay to Transmit Data on somebody else's Network. You've Turned Thailand into Slums to Make Your Satellite Broadband Networks. You Need Lots of Kerosine for your Top Gun Airforce (Jet engines) - You've Got Lots of Petrol/Diesel for Electricity And Piston Engine Aircraft. Such as a Cessna C172 or a Diamond DA42. Your Satellites are in The Clouds: Between 6,000 and 10,000 Feet. Piston Diesel Engines Work at that Altitude. You Need Kerosine for 15,000 Feet and above Jet Engines. The Difficulty is Building The Network. You Have To Pay to Transmit Data on somebody else's Network. You've Turned Thailand into Slums to Make Your Satellite Broadband Networks.