Bazel is seriously underappreciated. Clone of Google’s internal Blaze, it’s cross-language, battle-tested, and extensible. Decades of Google engineering expertise.
My favorite part: aggressive caching. I run all tests; it reruns only what changed, tracked by hashes. I’ve never seen it miss a change.
2.5s rebuild + test.
Earlier today I leaked AWS credentials to the world; except they weren’t real.
This is part of our launch for Honey Tokens (HT) at @infisical - a new class of fake credentials that can be used to trick attackers into thinking that they’ve stolen your real secrets.
HTs are useful as decoys for detecting bad actors and breaches in the event that they do happen. Under the hood, HTs are real AWS IAM credentials, except wired up to Infisical, but with zero permissions. When an attacker tries to use a HT, we notify you so you can stay proactive about further securing or rotating your real secrets.
In a world where credential breaches are becoming more common, we hope to give you all the tools needed to combat modern security threats.
More on this below 👇
@steeve Hahaha the X million dollars worth of machines we have racked in our colocation facility are now worth over twice as much as we paid for them thanks to all that ECC RAM.
Incremental builds deserve an incremental cache. Major props to Tyler French from the @buildbuddy team for landing this change in @bazelbuild that reduces cache uploads and disk cache size by 40%+ https://t.co/kwFRLMpMew
I am using @bazelbuild to download a Zig nightly build
To then build the same Zig version from source (we have patches)
Using a custom LLVM source repo
Which is also built from source
And all of this using remote execution with @buildbuddy.
@siggi@buildbuddy Exactly ! I show everyone I can how I’m pretty much alone doing codex cloud at this scale just because I don’t care containers are only 2 cores. While every other agents wait minutes doing things locally, mine loop at light speed while everything compiles and test w/ @buildbuddy