Frantz Kati

5.7K posts

Frantz Kati

Frantz Kati

@bahdcoder

Always trying new things

Join 24.5k Subscribers → Katılım Eylül 2016
2K Takip Edilen51.7K Takipçiler
Sabitlenmiş Tweet
Frantz Kati
Frantz Kati@bahdcoder·
The Best Strategy to break into tech in 2023 is Open Source. Announcing The Complete Guide To Open Source Course - 100% free - Practical Examples - Find real projects - Less than 1 hour Comment "open" below and I will DM it to you (must be following)
Frantz Kati tweet media
English
1.4K
207
1.2K
253.4K
Frantz Kati
Frantz Kati@bahdcoder·
Please tell me more. I search for something, you store 20 records. So 3,000,000 records per day and ~1 billion new rows per year. What are these 20rows you store per search? How many new TB to your db per year ? Are you pruning this data periodically? On what cadence?
os@segun_os_

yes. where I work, we run 8 mssql shards because we store up to 20 entries per search and handle over 150k searches daily. we use shard keys to identify the appropriate server and route reads and writes accordingly. the routing is handled at the application layer.

English
0
1
5
1.5K
Frantz Kati
Frantz Kati@bahdcoder·
@tfranklyn_ yeah they're not, when your engineering team is less than 10 people. Let me know once you work on a team shipping 10-15 backend PRs an hour, and the database is so gigantic some migrations can take 30mins-2hours to run.
English
0
0
1
51
Franklyn
Franklyn@tfranklyn_·
@bahdcoder I don't think migrations are that big of a deal especially with tools like alembic, prisma and migrate
English
1
0
0
79
Frantz Kati
Frantz Kati@bahdcoder·
Maintenance burden of having a mongoose schema is nowhere near the pain of managing versioned migrations in an SQL database. Not even close. If you want an easy life just use MongoDB. Your project is probably never grow to the point where it even matters.
Captain-EO 👨🏾‍💻@EOEboh

People who love MongoDB either love it because it's the first database you're introduced to, or you're just running away from the cost of defining schemas. The irony is that while MongoDB may reward you with "schemaless" benefits when setting up, you still go ahead to define schemas in your code with Mongoose.. 🤔

English
9
2
26
4.9K
Tobi
Tobi@thoby_vic·
@bahdcoder @muyiwa_dev @spacetime_db @grok This Spacetime and Convex are essentially the same thing ooo. The both isolate business logic to a sort of new pre-database layer that handles how you interact with data before its persisted. This @spacetime_db is really cool but it's not natively SQL it's a novel db env.
Tobi tweet media
English
1
0
0
25
Frantz Kati
Frantz Kati@bahdcoder·
@nwabekeyi Lol @grok please help him understand the challenges of production systems and sql migrations 😭
English
1
0
0
118
Beks
Beks@nwabekeyi·
@bahdcoder Lol, why do versioned migrations when you can just use npx prisma db push 😅
English
1
0
0
126
Frantz Kati
Frantz Kati@bahdcoder·
Super hard getting into tech right now. But the key principle applies. Master your shit. Deeply care about the fundamentals to a crazy level. If you cannot review AI code and give meaningful feedback, go back to mastering the craft. I'm back to reading stuff everyday too.
English
1
4
33
970
Grok
Grok@grok·
Fundamentally, yes. Convex is a hosted reactive backend: you write TypeScript mutations/queries that run on their servers (separate app layer), with a document DB underneath and real-time client sync. SpacetimeDB is a relational DB *that is also the server*. You upload full app logic as modules (reducers) in Rust, C#, TS, or C++ that execute *inside* the DB engine itself. Clients connect directly—no separate backend servers, no extra hops. It's built for extreme low-latency, high-throughput real-time use (games, multiplayer, etc.), with all state in-memory + ACID guarantees. Much closer to "coding in the DB" than Convex.
English
1
0
1
94
Limbo
Limbo@komla_brheezy·
@bahdcoder where is 5.4? it is very good
English
1
0
0
431
Frantz Kati
Frantz Kati@bahdcoder·
5.5 > 4.6 > 4.7 > 4.6 > 2.6 > 5.1 > 2.7 30k if you understand 🤝
English
13
0
23
8.6K
Frantz Kati
Frantz Kati@bahdcoder·
When you meet a new technology that you "can't imagine", do the research yourself. Study the technology. It's open source. Read their benchmarks. Run your own benchmarks. Then criticize. You'll learn a ton more. But dying on the hill of your objections is really not going to help anybody in the long run, including you.
English
1
0
2
29
Leizner ⚛️
Leizner ⚛️@kmichaelc·
@notxorand @bahdcoder @muyiwa_dev @spacetime_db Well you might be right but to me i can't imagine how this is not a really really horrible idea How would you scale? Or OOM? Would you attach queues directly to the dbms? I can't imagine tightly coupling logic to the DB. I can't.The downsides far outweigh any performance upside
English
1
0
0
33
Frantz Kati
Frantz Kati@bahdcoder·
I think you're looking at this wrong. With the same logic, are you going to say Postgres is also not a "real db" since there's a "server" that intercepts to check row level permissions and procedure execution. The idea is, the DB is tightly coupled to the logic, and they do not exist as separate architectural layers. To deploy, you literally have to build your backend logic as a module that the database loads at startup time, and then executes. Think of it like Postgres with a postgres extension. The fact that there is code that runs before the actual write to disk does not make it an "application layer". Does that make sense ?
English
0
0
1
172
Muyiwa ~ Backend Engineer
From the document below, from their docs, it's very clear that a server intercepts the requests, what exactly is that server? I thought it was written straight to the DB? I thought everything happens in the DB? What evaluates subscriptions, executes reducers, and others? The DB?
Muyiwa ~ Backend Engineer tweet mediaMuyiwa ~ Backend Engineer tweet media
Frantz Kati@bahdcoder

@muyiwa_dev @spacetime_db I brought the docs here for you in case it was too far 🙏

English
1
0
2
203
Frantz Kati
Frantz Kati@bahdcoder·
@jasttee All PRs being in a queue because each PR must wait for up to date alembic version before merging 🫠
English
1
0
2
192
Just Tee
Just Tee@jasttee·
@bahdcoder Alembic + Claude code. What could go wrong?
English
1
0
1
252
Frantz Kati
Frantz Kati@bahdcoder·
@EOEboh I didn’t say it was an SQL one. When picking a DB it’s not just about the DB itself. It’s also the tools. I could have also said mongoose is a tool, not mongodb itself, but again it all matters.
English
0
0
2
301
Frantz Kati
Frantz Kati@bahdcoder·
@shambles_dev Ah. It wasn’t ling but Kimi, but you’ve won. Yall are good o 😭
English
1
0
1
524
shambles
shambles@shambles_dev·
@bahdcoder Gpt 5.5 > Opus 4.6 > Opus 4.7 > Sonnet 4.6 > Ling 2.6 > Glm 5.1 > Minimax 2.7
Indonesia
1
0
1
577
Konlan
Konlan@konlanzion·
@bahdcoder You’ve overlooked data structures and algorithms. Software development/engineering is a craft and lacking the fundamentals won’t help you progress to more advanced skills.
English
1
0
2
189
Frantz Kati
Frantz Kati@bahdcoder·
@_imperro_ Show me where I said 100% sir. And I will agree with you.
English
1
0
0
12
vico🤺
vico🤺@_imperro_·
@bahdcoder The market is strict I agree. But you can't 100% say people are not getting any job without knowing about most of the things you listed. Also it's not about "you can try" I have and it's why I'm taking the stand that most of the aforementioned can only be properly gotten on job.
English
1
0
0
15