leo_e

7 ms·

1.
Show HN: Magpie – Fight AI sycophancy in code review with multi-model debate(github.com)
1 points by leo_e 7mo ago|0 comments
2.
Show HN: Magpie – I built a CLI where AIs argue about my code
2 points by leo_e 8mo ago|0 comments
3.
by leo_e 9mo ago
I think they are for different scales and scenario complexities.
4.
by leo_e 10mo ago
This will inevitably be used as ammunition against sideloading, but it’s really a lesson in supply chain trust. When we move away from walled gardens (which I support), the burden of verifying the "chain of custody" shifts to the
5.
by leo_e 10mo ago
The value here isn't just indexing; it's narrative. Reading the Linux kernel linearly is impossible because it’s not just code anymore—it’s 30 years of hardware quirks, scheduling theory, and architectural compromises solidified i
6.
by leo_e 10mo ago
That’s the truth
7.
by leo_e 10mo ago
The "SQLite doesn't scale" argument is usually just premature optimization masquerading as architectural wisdom. Unless you are actively hitting WAL contention limits (which is surprisingly hard to do on modern NVMe), the ope
8.
by leo_e 10mo ago
The choice of MIT for a kernel feels like setting up the project to be cannibalized rather than contributed to. We've seen this movie before with the BSDs. Hardware vendors love permissive licenses because they can fork, add their prop
9.
by leo_e 10mo ago
Forget the interaction layer. The moat isn't UI, it's Reliability Engineering. Right now, Manus and others are great at the 'Happy Path'. But when a 3-hour multi-step task hits a 503 error on step 47, does the agent grac
10.
by leo_e 10mo ago
Impressive numbers on paper, but looking at their site, this feels dangerously close to vaporware. The bottleneck for inference right now isn't just raw FLOPS or even memory bandwidth—it's the compiler stack. The graveyard of AI h
11.
Tell HN: Gemini 3 found a stack smash in a hex dump that I missed
8 points by leo_e 10mo ago|0 comments
12.
by leo_e 10mo ago
I'm willing to pay the "performance tax" of the web stack/self-hosting if it means my design files aren't held hostage in a proprietary cloud silo. Figma is fantastic software, but it has become a single point of fa
13.
by leo_e 10mo ago
It really puts our current definition of "latency" into a painful perspective. We have a machine running on 1970s hardware, a light-day away, that arguably maintains a more reliable command-response loop relative to its constraint
14.
Palo Alto Networks to Acquire Chronosphere (Creators of M3DB)(paloaltonetworks.com)
3 points by leo_e 10mo ago|0 comments
15.
by leo_e 10mo ago
To see this just as a hosting switch misses the bigger picture. This is the logical infrastructure conclusion of Zig's 'Zero Dependency' philosophy. Zig spent years removing dependencies on the system C compiler (zig cc), rem
16.
by leo_e 10mo ago
This trend is the absolute bane of early-stage startups. When you are bootstrapping and flying a team to a conference, sharing twin rooms is standard procedure to stretch the runway. There is nothing that kills the vibe of a "strategic
17.
by leo_e 10mo ago
I honestly struggled to tell if the top comment on that thread was generated or real. It captured the specific type of "dismissive pedantry" we see here so perfectly. It makes you wonder: if a 70B parameter model can perfectly sim
18.
by leo_e 10mo ago
The most concerning part isn't the vulnerability itself, but Google classifying it as a "Known Issue" ineligible for rewards. It implies this is an architectural choice, not a bug. They are effectively admitting that you can&
19.
by leo_e 10mo ago
As someone fighting the C++ toolchain daily, there is a painful irony in seeing APT—the tool supposed to solve dependency hell—creating its own dependency crisis. I sympathize with the maintainers of retro hardware. But honestly? Holding ba
20.
by leo_e 10mo ago
We learned this the hard way with "cold" backups stored in a literal safe. We treated NVMe drives like digital stone tablets. A year later, we tried to restore a critical snapshot and checksums failed everywhere. We now have a pol
21.
by leo_e 10mo ago
Papers like this are fascinating engineering, but dangerous marketing. They convince every Series A startup that they need a multi-region federated control plane for their 50 microservices. I spend half my time convincing my team not to emu
22.
Are You Sure You Want to Use MMAP in Your Database Management System? (2022) [pdf](db.cs.cmu.edu)
5 points by leo_e 10mo ago|0 comments
23.
by leo_e 10mo ago
Hardware is faster, but the "abstraction tax" is higher than ever. As someone currently fighting to shave megabytes off a C++ engine, it hurts my soul to see a simple chat app (Electron) consume 800MB just to idle. We spent the la
24.
by leo_e 10mo ago
Hard agree. The hidden cost of 'cheap' models is the complexity of the retry logic you have to write around them. If a cheaper model hallucinates halfway through a multi-step agent workflow, I burn more tokens on verification and
25.
by leo_e 10mo ago
It is honestly refreshing to see constraints like this again. In my cloud infrastructure work (C++), we have gotten lazy. We bloat our containers because 'RAM is cheap'. Seeing a system designed to fit into 1MB reminds me that per
26.
by leo_e 10mo ago
As someone dealing with open-source compliance in distributed systems, the purity tests in these comments are exhausting. Hardware is messy. IP licensing for sensors and radios is a nightmare. Getting a functional OS out with "only&quo
27.
Ask HN: Is starting a new project in C++ just a mistake in 2025?
4 points by leo_e 10mo ago|3 comments
28.
by leo_e 10mo ago
The trouble with mmap is the performance cliff. A node goes from 'fine' to 'dead' almost instantly, which breaks our balancing logic. You are right that we need better backpressure. Instead of a smarter coordinator, we p
29.
by leo_e 10mo ago
You're right. O_DIRECT is the endgame, but that's a full engine rewrite for us. We're trying to stabilize the current architecture first. The complexity of hidden page fault blocking is definitely what's killing us, but
30.
by leo_e 10mo ago
I admit PSI wasn't on our radar for this specific issue. We've been staring at RSS and page fault counters, but they are indeed too noisy in an mmap-heavy workload. Checking /proc/pressure/memory to distinguish betw

More ›