Hacker News: dikei

New comment by dikei in "AWS multiple services outage in us-east-1"

dikei — Mon, 20 Oct 2025 10:30:20 +0000

Some global services have control plane located only in `us-east-1`, without which they become read-only at best, or even fail outright.

https://docs.aws.amazon.com/whitepapers/latest/aws-fault-iso...

New comment by dikei in "Fire destroys S. Korean government's cloud storage system, no backups available"

dikei — Mon, 06 Oct 2025 10:22:09 +0000

For paper documents, you'd make at least a few copies for storage at the source, and then every receiver will get his/her own notarized copies.

Electronically, everyone just receives a link to read the document.

New comment by dikei in "ICPC 2025 World Finals Results"

dikei — Fri, 05 Sep 2025 09:30:45 +0000

Start training them from a very young age: competitive coders typical start in secondary/high school or even earlier.

Russia has specialized schools to support this kind of education, so are China and many other countries. Thus, they rank very high in IOI and ICPC contests

New comment by dikei in "Show HN: OctaneDB – Fast, Open-Source Vector Database for Python"

dikei — Sat, 23 Aug 2025 10:51:31 +0000

Big claims with not a single benchmark to back them up

New comment by dikei in "7-Zip for Windows can now use more than 64 CPU threads for compression"

dikei — Wed, 09 Jul 2025 11:51:24 +0000

Well, only a lunatic would use ZIP with anything but DEFLATE/DEFLATE64

New comment by dikei in "7-Zip for Windows can now use more than 64 CPU threads for compression"

dikei — Wed, 09 Jul 2025 11:14:51 +0000

I use ZSTD a ton in my programming work where efficiency matters.

But for sharing files with other people, ZIP is still king. Even 7z or RAR is niche. Everyone can open a ZIP file, and they don't really care if the file is a few MBs bigger.

New comment by dikei in "Ticket-Driven Development: The Fastest Way to Go Nowhere"

dikei — Thu, 26 Jun 2025 12:09:56 +0000

I hate ticket-driven development, but clearing ticket is one way to stay sane when the final objective looks crazily out-of-touch.

New comment by dikei in "Better Auth, by a self-taught Ethiopian dev, raises $5M from Peak XV, YC"

dikei — Thu, 26 Jun 2025 05:52:59 +0000

Yeah, and all the popular web frameworks include authn and authz as a core component.

New comment by dikei in "How Cloudflare blocked a monumental 7.3 Tbps DDoS attack"

dikei — Wed, 25 Jun 2025 02:56:10 +0000

OpenWRT stopped doing this 10 years ago, as it was too much hassle to pick a drink that satisfy everyone.

New comment by dikei in "Jemalloc Postmortem"

dikei — Fri, 13 Jun 2025 03:04:14 +0000

I still remember the day when I used jemalloc debug features to triage and resolve some nasty memory bloat issues in our code that use RockDB.

Good times.

New comment by dikei in "A New ASN.1 API for Python"

dikei — Sun, 20 Apr 2025 03:47:12 +0000

The difference is the quality of the OSS implementation: most OSS ASN.1 tool choke on the enormous 3GPP specs and others used in the telco industry, thus cannot generate 100% valid code.

For some use-cases, you can get by with manually adjust the generated code. That works until the hardware vendors release a new device that use a more modern 3GPP specs and your code start breaking again.

When using a commercial ASN.1 tooling, they often update their compilers to support the latest 3GPP specs even before the hardware vendors, and thus supporting a new device is way simpler.

New comment by dikei in "A New ASN.1 API for Python"

dikei — Fri, 18 Apr 2025 15:09:39 +0000

DER is still easy, UPER (unaligned packed encoding rules) is so much harder, yet it's prevalent in Telecom industry. Last I checked, there was no freely available tool than can handle UPER l00%

New comment by dikei in "We are the "thin blue line" that is trying to keep the code high quality"

dikei — Fri, 14 Feb 2025 03:14:58 +0000

It's not about the VCS, it's about API compatibility being broken.

If your code is out-of-tree, every time someone changes an API that break your build, it's your responsibility to update your for and fix it.

If your code is mainline, the people who change API and break the compilation are on the hook to fix it.

New comment by dikei in "OpenWrt 24.10.0 – First Stable Release"

dikei — Thu, 06 Feb 2025 10:52:31 +0000

The use-case is because it's so cheap, why bother with anything less.

Currently, I can get 1Gbps Internet for $15, while the cheapest package is 200Mbps for $5. I expect they'll offer 10Gbps in my area in the next few years for the same cost as the 1Gbps now.

Still, at that speed, the router CPU can actually become the bottleneck, and OpenWRT currently has pretty poor support for hardware accelerated routing.

New comment by dikei in "The Canva outage: another tale of saturation and resilience"

dikei — Mon, 13 Jan 2025 16:13:44 +0000

One reason for Spotify's move away from p2p was it was absolutely a no-go on mobile platform, which was rapidly becoming dominant at the time.

New comment by dikei in "Show HN: LogLayer – Unified logger that routes logs to various logging libraries"

dikei — Wed, 08 Jan 2025 05:11:04 +0000

I prefer Log4J2 built-in JSON Template Layout to Logback's various JSON output solution.

The API of Log4J2 and SLF4J v2 are similar enough that it does not matter for basic usage, I don't see the benefit of SLF4J in a project that's already using Log4J2.

New comment by dikei in "Show HN: LogLayer – Unified logger that routes logs to various logging libraries"

dikei — Mon, 06 Jan 2025 07:09:28 +0000

Not a good example, as the de-facto logging interface for Java was SLF4J.

Though I myself prefer Log4J2, as I'm already using Log4J2 as the concrete logging implementation.

New comment by dikei in "Kuvasz-streamer: open-source CDC for Postgres for low latency replication"

dikei — Fri, 03 Jan 2025 11:58:10 +0000

We use the same setup, though we use PGBouncer so after switching primary we just force reconnect all clients from PGBouncer instead.

The clients will have to retry on-going transactions, but that's a basic fault tolerant requirement anyway.

New comment by dikei in "Kuvasz-streamer: open-source CDC for Postgres for low latency replication"

dikei — Fri, 03 Jan 2025 11:34:56 +0000

I'd be more similar to Debezium-server that runs everything in the same process, than regular KafkaConnect-based Debezium.

However, this only does postgres-postgres, so it's a lot more limited compared to Debezium.

New comment by dikei in "Electric (Postgres sync engine) beta release"

dikei — Thu, 12 Dec 2024 09:13:05 +0000

Ok, so PGLite is used in the Firebase Data Connect's emulator: a test utility, not even the client library.

It's a bit disingenuous to lump it in the same sentence and try to mislead user that Google's adopting Electric.