HNHacker News
TopNewBestAskShowJobs

rldjbpin

298 karma · joined August 2, 2022

submissionscomments
rldjbpin··on Dependency cooldowns turn you into a free-rider
The moral judgement of a practice not unknown to those who handle production deployment (takes me back to the days where i had to use a local maven repository with dated dependencies) is on a very shaky foundation.

We used to focus more on finding issues before a new release, and while it remains common to find bugs in older ones, not having enough users should not be used as a crutch for testing.

> (dependency cooldowns) don't address the core issue: publishing and distribution are different things and it's not clear why they have to be coupled together.

Besides some edge cases for a large project, the core issue remains code quality and maintainability practices. The rush to push several patches per day is insane to me, especially in current AI ecosystem.

Breaking changes used to have enough transitionary period, see Python 2 to 3, while today it is done on a whim, even by SaaS folks who should provide better DX for their customers. Regardless, open-source/source-available projects now expect more from their users, and I wonder how much of it remains reasonable.

rldjbpin··on Apple's accidental moat: How the "AI Loser" may end up winning
the core thesis of the article and comparison with non-"losers" seem vague at best. comparing model providers with hardware vendor with software ecosystem is like differentiating between apple to oranges.

if hardware moat was to be discussed, then compare with nvidia, amd and google's tpu division perhaps. in-house intelligence is best left alone for apple. they are relying on the "peers" for underlying capabilities as is. [1] [2]

outside of inference and (pro/con)sumer space, there is little to offer for the enterprise or the people developing the lowest end of the stack. even the recent tinygrad egpu is shockingly slow [3]. which might made gb10 look much more capable for in-house training.

regardless, most of the industry "moat" does not appear sustainable at best. only time will tell how it will turn out for everyone but on a positive note, apple does not put all its eggs in this basket, which is probably wiser.

[1] https://news.ycombinator.com/item?id=40636980

[2] https://news.ycombinator.com/item?id=46589675

[3] https://www.youtube.com/watch?v=C4KWsmezXm4

rldjbpin··on Apple's accidental moat: How the "AI Loser" may end up winning
> If a couple more iterations of this, say gemma6 is as good as current opus and runs completely locally on a Mac, I won’t really bother with the cloud models.

> That’s a problem.

While improvements should continue rolling in, and might even match current SOTA in benchmarks down the line, is it "good enough"?

Hard to believe we have reached that stage with current models, which would continue to stretch beyond what we can economically run. Call it skill issue, or try to fix it with a revolutionary harness, it seemingly takes a village to get it all working. Maybe by then we will have good enough ecosystem in these layers too, but if current capabilities is the benchmark, it might need more time in the oven.

rldjbpin··on Google removes "Doki Doki Literature Club" from Google Play
weird situation regardless. this came out like a decade ago, and while there is a serious undertone, this has been widely discussed during the original virality period.

funnily enough one of the largest Youtubers made a gameplay video of the desktop game [1]. don't believe anything was modified when the port was made. hope this gets resolved just like the recent wireshark fiasco.

[1] https://www.youtube.com/playlist?list=PLYH8WvNV1YEnYtfmH-sGR...

rldjbpin··on I run multiple $10K MRR companies on a $20/month tech stack
might also consider using a solution such as disco [1] which avoids requiring VMs while still allowing the ergonomics of cloud services.

at the end of the day, for a "$20/month" stack, there may probably not be enough services running to make sense for going dedicated yet.

[1] https://github.com/letsdiscodev/

rldjbpin··on South Korea introduces universal basic mobile data access
this does not come across that impressive beyond the surface. however, given the market inefficiencies (see Twitch saga over double charging of bandwidth [1]) despite having some of the fastest and (relatively) affordable internet has always made it an interesting case study.

throttled speed beyond cap is something i've grown up seeing since the ADSL days, but mandating it across the handful of providers can perhaps help with the odd Line text or two.

[1] https://news.ycombinator.com/item?id=38539167

rldjbpin··on We've raised $17M to build what comes after Git
building the future of source control over the incumbent tech does not bring much excitement personally.

does not takeaway the interest and funding they are able to bring. raising money is not always a bad thing, if say, there will be a webapp to go alongside their approach to workflows.

rldjbpin··on Filing the corners off my MacBooks
misread two 'l's instead of one and imagined it would be about the rounded corners in the latest macos.

personally, having recently switched to one, the edges were a bother at first. but over the months grew to tolerate them just like callouses. opening the lid from the sides make me appreciate the need for a lip. however, i really hope they refine their approach for the rumored new model with better tooling and more comfortable.

rldjbpin··on MegaTrain: Full Precision Training of 100B+ Parameter LLMs on a Single GPU
even if it scales down, the host-to-gpu memory ratio is around 10 to 1 (1.5 TB vs 141 GB) [1].

might need some odd system builds, improbable with current pricing, to replicate or better the ratio. such as a ~256 GB system for a 4090 (with 24 GB VRAM).

[1] https://www.nvidia.com/en-us/data-center/h200/

rldjbpin··on Apple approves driver that lets Nvidia eGPUs work with Arm Macs
based on your card, it should be using a decent bit of the 600w or so passed through the new 16-pin connector. goes without saying, a proper PSU (doesn't have to be ATX, but at least 750W to be on the safer side) is a must.

for thunderbolt enclosures, consider going through the list - https://egpu.io/best-egpu-buyers-guide/#tb3-enclosures

zero idea about mac support so YMMV.

rldjbpin··on LinkedIn uses 2.4 GB RAM across two tabs
on my device (running chrome with one extension), this page is consuming ~100 MB.

not to takeaway from the obviously bloated website discussed here, but i wonder who much of this is an overhead from the browser than the page bloat itself.

rldjbpin··on AMD's Ryzen 9 9950X3D2 Dual Edition crams 208MB of cache into a single chip
or an older version of puppy linux?
rldjbpin··on $500 GPU outperforms Claude Sonnet on coding benchmarks
> coding benchmarks

> V3 phases were designed and tuned for LiveCodeBench.

only compared on the above benchmark, while this has been identified and being improved for the next version.

curious to see how it compares across the board against the base model (Qwen3-14B-Q4_K_M)

rldjbpin··on Hold on to Your Hardware
while one should make the most of our electronics, the theory behind the article's headline is fuelled by fear-mongering most of us are jaded by.

personally been stretching the most life out of my machines. particularly because there has been some supply-demand "issue" happening ever since i could afford to.

my previous computer was a laptop without any tensor cores. it was bought right on the onset of lockdown-related supply chain issues. i was ok to stretch it beyond the better half of a decade before it died on me.

now had to build a computer out of necessity while paying 5x for ram. it took a lot to suppress the internal rationalisation of waiting it out for overall costs to improve. but in this decade, they almost never have!!

stepping back, we forget to appreciate how the cost of computing against cost of living have been so good to us for almost two decades. now thanks to opportunistic greed and real issues, it is coming back to the older times. one could still, for the price of a month's rent*, afford a decent computer which can last years.

rldjbpin··on Mistral AI Releases Forge
> Not everyone is obsessed with code generation.

They (still) are. https://news.ycombinator.com/item?id=47404796

rldjbpin··on How kernel anti-cheats work
> pretty much the only company that could make this work is Valve

at least when focusing on counter-strike (CSGO/CS2), they've tried tons of ways to segregate the player base in terms of trustworthy vs not.

from "anti-cheat" vs not, verifying users using phones, paying vs not, you name it.

none of their initiatives managed to ward off the bad actors from the "secured" version. does not give me the confidence that they could make a system work effectively, but something that can work cross-platform, perhaps.

rldjbpin··on How kernel anti-cheats work
> What does make cheat developers afraid is AI

there are already systems that just use the video feed of the game and provide external inputs for popular games (see [1] for a goofy demo). this goes beyond the pc where the game is being run.

kernel-level anti-cheats came at the wrong standpoint imho, and since there are still cheats in games that support it (e.g. EA FC 25, some CoD Warzone stories, etc.), not quite worth the compromise.

[1] https://www.youtube.com/watch?v=T4rrcw_oRVs

rldjbpin··on What makes Intel Optane stand out (2023)
the tech had so much potential, but it was during the era when Intel dominated. other than the pcie add-in cards that can act has fast cache, all the major use cases were down to their whims. platform support (or the lack thereof) killed it before it could be realized.

too bad there was no apple moment (e.g. with thunderbolt) where a third-party was allowed to make it more widespread.

rldjbpin··on Games with loot boxes to get minimum 16 age rating across Europe
to the dozens who use the rating system as a deciding factor for getting games for their kids, i suppose this will help.

meanwhile, one of the game that would have been affected by it, Counter-Strike 2, is already rated M by ESRB [1]. it is undergoing a major case in NY as we speak, and there are many professional players, also recognized by the devs, that openly stated they played the game since their early teens. [2]

it does mean that a lot of more suitable games for younger audience, such as the sports title released every year. but a lot of them already have free titles with pay-to-win mechanics. i wonder if the enforcement would really differ any more than it currently is.

[1] https://www.esrb.org/ratings/9406/counter-strike/

[2] https://www.hltv.org/player/19230/m0NESY

rldjbpin··on Parallels confirms MacBook Neo can run Windows in a virtual machine
buying a $500 laptop to run a software that costs starting from $99 a year* is a funny situation. anyways, the fact that it works means in terms of architecture the A and M series do not need much work between them, or it has been done already.

with the (premium) chromebook parallels being drawn, having the linux experience, a la chroot, would be a more interesting point for the crowd reading this.

rldjbpin··on Mouser: An open source alternative to Logi-Plus mouse software
depending on your os*, you may have a better experience with a first-party lightweight tool called onboard memory manager [1].

single binary, no install nor admin rights needed. let's you change keybinds and other mouse features.

*might be Windows-only

[1] https://support.logi.com/hc/en-gb/articles/360059641133-Onbo...

rldjbpin··on A decade of Docker containers
use the tech at work on a daily, but glad OCI exists to not be fleeced by the enterprise arm.

installing docker desktop on my personal laptop permanently bricked it, right in the middle of chip/memory shortage. thanks docker!!!

rldjbpin··on RAM now represents 35 percent of bill of materials for HP PCs
for those shopping for new PCs and are open to build one, it seems like the overall BOM has been overall lesser than a few months back. the big caveat being that you need a decent GPU.

having been in the market for one, i did make some compromise for the build (single stick of 16 GB for now; a non "future-proof" GPU within the budget). however for a decent spec (last-gen x3D CPU, mid-range RTX) build, the GPU price reduction made up for the premium on RAM.

the sad reality since the turn of the decade is that the $1000 mark for a well-rounded (gaming) system has now bumped up to $1500-$2000.

crazy time to be alive, where on laptop side, macs are now a "decent" value. especially if you were going to get the higher spec to unlock the specific memory tier. thanks work for setting me up well with one!

rldjbpin··on Coding agents have replaced every framework I used
so many parties peddling AI related services should also agree with the sentiment (so do i). a lot of people in the stack are acting as wrappers. most of their tools are always changing (e.g. see the commit counts in openclaw project), often full of breaking changes and while forcing to migrate to newer version regularly.

if you want something specific without ever-changing needs, it makes sense to do it with fewer dependencies as possible. this should be even more ideal if you believe the code made is of high quality, as it reduces the pain of updating the project for non-functional reasons.

this sentiment ironically goes against the existence of these services, as they themselves are always working on matching the ever-changing dependencies to give the impression of a stable platform.

only if it was possible to have the cake and eat it too

rldjbpin··on Microsoft account bugs locked me out of Notepad – Are thin clients ruining PCs?
> Windows 10 (LTSC) has been the best operating system experience of my life

Unless you've stumbled upon it by chance, the LTSC version of Windows is by far the recommended approach by forums online, particularly for those who do not want to run random scripts to remove unwanted elements.

Windows 11 happen to have its own variant [1], I wonder how it compares to the gold standard of the previous version.

[1] https://www.microsoft.com/en-us/evalcenter/evaluate-windows-...

rldjbpin··on CIA suddenly stops publishing, removes archives of The World Factbook
used it extensively in high school as a more approachable way to learn about countries without the daunting amounts that wikipedia pages tend to have.

come to realise down the line the "writer's biases" in the latter half of the articles. almost comical to see how nuanced situations are distilled down to "our perspective". recall how you could "learn" about the drug trade situation in almost any country, regardless of how you perceived the country's safeness before reading the page.

rldjbpin··on My AI Adoption Journey
not quite as technically rich as i came to expect from previous posts from op, but very insightful regardless.

not ashamed to say that i am between steps 2 and 3 in my personal workflow.

>Adopting a tool feels like work, and I do not want to put in the effort

all the different approaches floating online feel ephemeral to me. this, just like for different tools for the op, seem like a chore to adopt. i like the fomo mongering from the community does not help here, but in the end it is a matter of personal discovery to stick with what works for you.

rldjbpin··on See how many words you have written in Hacker News comments
0.000622 here :)

somehow almost close to the top 10k despite writing one-hundredth as much as the top guys.

rldjbpin··on I dumped Windows 11 for Linux, and you should too
> My laptop is a 2014 MacBook Air, which is still a pretty solid machine.

as a long-term contributor to a website that benchmarks (quite extensively!) notebooks of all sorts, quite interesting to see the choice of portable machine.

quite aligned with the overall take and was well-supported with examples.

rldjbpin··on AWS raises GPU prices 15% on a Saturday, hopes you weren't paying attention
> Game publishers have already publicly floated the idea of not selling their games but charging per hour. Imagine how that impact Call of Duty or GTA.

Don't you worry! R* is already selling one (albeit optional) [1]. it is expected to be there from day 1 in their next title. perhaps this could become closer to reality once the current subscription becomes commonplace.

[1] https://www.rockstargames.com/gta-plus/benefits

← PreviousPage 5 of 25Next →