RT @ClickHouseDB: 📢 We're live at ClickHouse Open House in SF! Stay tuned for an action-packed day of exciting product updates, customer st….

RT @ClickHouseDB: 📣 New deep-dive: "Under the Hood: Building MySQL Change Data Capture in ClickPipes.". In this te….

RT @gwenshap: For years, my favorite trait in developers was curiosity. Curios developers learn fast and write more reliable code because….

RT @carlsverre: What if your app could combine the simplicity of physical replication with the efficiency of logical replication? Meet Graf….

Seeing a lot of discussions around sync engines lately, which motivated me to share some lessons learned from working on Livegraph in my early days at Figma. Hope you find it useful 😊.

RT @sriramsubram: 🔊 Announcing new opening at Nile. We are looking to hire a rockstar Auth engineer to work on Nile Auth. It would be great….

Very interesting discussions in this YC thread -- and yes, the trend of building stuff on object storage continues!.

RT @aliasaria: A long time ago, there was a secret meeting of the top minds at BlackBerry. Their job was to save the company over a single….

😂😂😂.

This is a practical and honest writeup of working on developer infrastructure at scale: .

My head hurts thinking about durable execution for too long, but thank god my brain has the durable execution feature.

Just published my first Substack blogpost: Demystifying Distributed Checkpointing.

Kudos to folks that write weekly newsletters. Not sure where y’all find the time. 😂.

It’s been a while since I wrote a blogpost. Definitely underestimated the amount of effort and time but I’m getting close to wrapping up and publishing! Stay tuned!.

Anyways, I am a Bazel fan but it’s hard similar to writing functional code without side-effects is hard when it comes to real-life programming. It gives you back as much as you are willing to put into it.

Other criteria to even consider Bazel at all, IMHO:.- monorepo.- polyglot codebase.- more than a couple hundred engineers.- build has become the bottleneck to productivity.

If your company doesn’t have the resources to set up a Bazel platform team, then your builds are probably not that slow, or there are other low hanging fruits to improve build speed, and so you are better off delay your Bazel migration journey.

Not to downplay the need of Bazel knowledge sharing — however it’s not actually practical for all engineers (especially product engineers) to learns the ins and outs of Bazel. If product engineers are writing custom rules then the company is not investing in it enough.

One lesson learned from working on Bazel migration at Figma is that having a centralized Bazel platform team to build deep expertise is crucial to the success of Bazel adoption.

I’m picturing a future where most of the things are built by AI but we got a small number of tools and sites advertised with “build by human, build for human” or “artisan software, hand-typed with love”.