Alan Floyd

19 | Geology undergrad → aspiring paleontologist | Hobbyist coder, learning Python/Linux/MySQL/Git | vibe coding | random daily stuff

[Alan Floyd @alanfloyd_]

Hello, world! 👋

I’m Alan, a geology undergraduate from China, and I hope to pursue a career in paleontology in the future.

My main interests include:
Computer Science | Paleontology | Rock Music | Reading

Tech stack:
Python for data analysis | MySQL | Linux | C | Currently learning Rust

For personal projects, I also use AI-assisted development with Vue 3, TypeScript, and C#.

GitHub: https://t.co/cn09KfAcgw
Personal site: https://t.co/yGVLWGgdAp

My English is still a work in progress, and I sometimes use translation tools, so I appreciate your understanding.

Glad to meet you all! ( ´ ▽ ` )ﾉ

[Alan Floyd @alanfloyd_]

@stejas809 Hi, let’s connect

[Alan Floyd @alanfloyd_]

@RealNipuna Followed you, let’s connect 👋

[Alan Floyd @alanfloyd_]

@elbog29 Hi,followed you,let’s connect

[Alan Floyd @alanfloyd_]

@nakularora Hi,followed you,let’s connect

[Alan Floyd @alanfloyd_]

@ydhshhdsj Hi,followed you,let’s connect

[Alan Floyd @alanfloyd_]

@iancccho Hi,followed you,glad to connect

[空言 @__soragoto__]

这么几年过去了，还有多少人会打开传统 IDE 呢

Jetbrains 作为老牌 IDE 大厂，之前发布过一个叫 Air 的产品，这玩意多少有点之前那个已经爆死的VScode 的竞品 (Fleet) 的影子，刚开始的时候我还去试用过一次，保留了他家那个非常好用的 git 视图，可惜非英文输入法有十分严重的 Bug，根本无法输入中文，日文等东西，现在我还没好好用过这玩意…

然后最近我又发现，他们居然还做了个叫 thinkrail 的玩意，他们说这玩意是基于 Pi coding agent 的 IDE，是个 Agent naitve IDE

他们称这玩意把 Git Worktree 当基本工作单位，能做到让不同 Agent 在独立的工作区并行干活，还搞了一套 Living Specs 机制，好让 Agent 能跨 session 继承知识……

我们的老牌 IDE 大厂现在一手抓把 Agent 集成进老 IDE，一手抓 Agent native IDE，不知道这两新玩意以后是又爆死一个还是两个东西合并，或者，两个一起爆死？

有没有勇士想去试试这个叫 ThinkRail 的玩意？

https://t.co/GDzMAH6Xpr

对了，如果你刚好是 Jetbrains 的老用户，而且很久没打开过他们家 IDE 了，也可以现在上去看看，不用下载这两新玩意，老的 IDE 更新一下，就能用上 IDE 内置的 MCP server 了，其实还挺好使来着 …

[Alan Floyd @alanfloyd_]

@Stringcodes Hey,let’s connect,followed you

[Alan Floyd @alanfloyd_]

@Ravi_Nileshbhai followed you,let’s connect

[Alan Floyd @alanfloyd_]

Hey, followed you,glad to connect!

I’m 19, a geology student from China, also into coding.Learning Python for data analysis, along with Linux and MySQL.

I’ve been working on a Windows desktop widget using Rust and Tauri lately, with the help of AI :)

Here’s the project if you’re interested!

https://t.co/Jx89juP1oY

[Alan Floyd @alanfloyd_]

@gauravvishgkp Congrats! Just followed you,let’s connect

[Alan Floyd @alanfloyd_]

Hey, just followed you! Glad to connect!

I’m 19, a geology student from China, and I’m also into coding and AI :)

Been building some stuff with the help of AI lately!

And I totally agree—language shouldn’t be a barrier. I really wanna meet people who share my interests and have some interesting conversations along the way :)

[Alan Floyd @alanfloyd_]

@dramaricic 👋, followed you, let’s connect

[Alan Floyd @alanfloyd_]

@juanpmd Hi, followed you, happy to connect

[Alan Floyd @alanfloyd_]

@mhaurel_ Followed you! Let’s connect

[Alan Floyd @alanfloyd_]

TBH, I’m not really interested in X’s creator rewards. I just wanna meet people who share my interests, have some genuine conversations, and share my experiences.

But as a new account with very few followers, my posts tend to get buried under an endless flood of follow-for-follow posts. So I ended up joining in for now, too ¯\_(ツ)_/¯

I guess I’ll stop once I’ve built up a decent following. These repetitive, almost robotic interactions get boring pretty quickly…

Still, whenever someone actually takes the time to have a real conversation with me, even if it’s just a few words, it makes the whole experience feel a little less dull.
:D

[lifcc @mylifcc]

给 PostgreSQL 做实时同步，终于不用在「装第三方插件污染源库」和「搭一套吃掉几 GB 内存的 Debezium 集群」之间二选一了。
PGRun 团队刚用 Rust 开源了 pgsync——一个把内存控制到极致的单二进制 PostgreSQL 实时逻辑复制（CDC）引擎。实测将 1GB 数据库做全量一致性快照时，峰值内存（RSS）仅 33MB；日常增量流式同步稳态仅 15MB，单机跑 20 个并发任务每进程才 9.5MB。
它最狠的不是快，而是死磕「绝对正确性（Correct before fast）」的三个架构设计：
1️⃣ 零侵入全量 + 增量无缝切换：基于 PG 原生 pgoutput 协议（源库无需装 wal2json 等任何扩展），先通过导出快照（Exported Snapshot）完成一致性全量拷贝，再精准从对应 LSN 位点切入实时 CDC 流。
2️⃣ 1:1 事务边界与原子 Checkpoint：源库每个事务在目标库严格作为唯一事务按序提交，且同步位点（pgsync.job）直接包在目标库同一个事务里落盘。哪怕遭遇 15 轮连续随机 SIGKILL 强杀，重启后依靠 LSN 幂等去重，数据 Hash 依然 100% 严丝合缝。
3️⃣ 22 项前置体检与零静默丢数：启动前一句 pgsync check 自动排查 22 项权限与配置隐患；同步中途一旦遇到目标库缺列或异常，立即显式停机报错，绝不静默丢弃一行数据。
你们目前在生产或测试环境做 PG 实时同步，主要用哪套方案？踩过最深的一个坑是什么？

[Alan Floyd @alanfloyd_]

Hey, followed you,glad to connect!

I’m 19, a geology student from China, also into coding.Learning Python for data analysis, along with Linux and MySQL.

I’ve been working on a Windows desktop widget using Rust and Tauri lately, with the help of AI :)

Here’s the project if you’re interested!

https://t.co/Jx89juP1oY

[Alan Floyd @alanfloyd_]

@anupamas02 @X Hey, followed you,glad to connect!

I’m 19, a geology student from China, also into coding and AI :)

I’ve been working on a Windows desktop widget using Rust and Tauri lately :)

Here’s the project if you’re interested!

https://t.co/MWW67icQSz