<rss version="2.0" xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:atom="http://www.w3.org/2005/Atom"><channel><title>Hacker News: pshirshov</title><link>https://news.ycombinator.com/user?id=pshirshov</link><description>Hacker News RSS</description><docs>https://hnrss.org/</docs><generator>hnrss v2.1.1</generator><lastBuildDate>Mon, 06 Apr 2026 07:42:17 +0000</lastBuildDate><atom:link href="https://hnrss.org/user?id=pshirshov" rel="self" type="application/rss+xml"></atom:link><item><title><![CDATA[New comment by pshirshov in "Running Gemma 4 locally with LM Studio's new headless CLI and Claude Code"]]></title><description><![CDATA[
<p>Only for chat sessions, not for agentic coding. It's just too slow to be practical (10 minutes to answer a simple question about a 2k LoC project - and that's with a 5070 addon card).</p>
]]></description><pubDate>Sun, 05 Apr 2026 23:29:26 +0000</pubDate><link>https://news.ycombinator.com/item?id=47655018</link><dc:creator>pshirshov</dc:creator><comments>https://news.ycombinator.com/item?id=47655018</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=47655018</guid></item><item><title><![CDATA[New comment by pshirshov in "Running Gemma 4 locally with LM Studio's new headless CLI and Claude Code"]]></title><description><![CDATA[
<p>Even tried gemma4:31b and gemma4:31b with 128k context (I have 72GiB VRAM). Nothing. I'm cursed I guess. That's ollama-rocm if that matters (I had weird bugs on Vulkan, maybe gemma misbehaves on radeons somehow?..).<p>UPD: tried ollama-vulkan. It works, gemma4:31b-it-q8_0 with 64k context!</p>
]]></description><pubDate>Sun, 05 Apr 2026 23:20:02 +0000</pubDate><link>https://news.ycombinator.com/item?id=47654953</link><dc:creator>pshirshov</dc:creator><comments>https://news.ycombinator.com/item?id=47654953</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=47654953</guid></item><item><title><![CDATA[New comment by pshirshov in "Running Gemma 4 locally with LM Studio's new headless CLI and Claude Code"]]></title><description><![CDATA[
<p>For some reason, that doesn't work for me, claude never returns from some ill loop. Nemotron, glm and qwen 3.5 work just fine, gemma - doesn't.</p>
]]></description><pubDate>Sun, 05 Apr 2026 20:58:22 +0000</pubDate><link>https://news.ycombinator.com/item?id=47653802</link><dc:creator>pshirshov</dc:creator><comments>https://news.ycombinator.com/item?id=47653802</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=47653802</guid></item><item><title><![CDATA[New comment by pshirshov in "Show HN: BreezePDF – Free, in-browser PDF editor"]]></title><description><![CDATA[
<p>> BentoPDF (12.3k stars): <a href="https://github.com/alam00000/bentopdf" rel="nofollow">https://github.com/alam00000/bentopdf</a><p>> PDFCraft (3.6k stars): <a href="https://github.com/PDFCraftTool/pdfcraft" rel="nofollow">https://github.com/PDFCraftTool/pdfcraft</a><p>> PDFLince (31 stars): <a href="https://github.com/GSiesto/pdflince" rel="nofollow">https://github.com/GSiesto/pdflince</a></p>
]]></description><pubDate>Sun, 29 Mar 2026 21:53:36 +0000</pubDate><link>https://news.ycombinator.com/item?id=47567788</link><dc:creator>pshirshov</dc:creator><comments>https://news.ycombinator.com/item?id=47567788</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=47567788</guid></item><item><title><![CDATA[New comment by pshirshov in "Show HN: BreezePDF – Free, in-browser PDF editor"]]></title><description><![CDATA[
<p>> Lots of options out there and everyone can choose for themselves.<p>You are shilling your stuff at a wrong place, I think. Better apply to YC or, I dunno, go public. Also add some nice catch phrases (e.g. "Blazing Fast", "Production Ready") and emojis here and there.</p>
]]></description><pubDate>Sun, 29 Mar 2026 21:11:10 +0000</pubDate><link>https://news.ycombinator.com/item?id=47567363</link><dc:creator>pshirshov</dc:creator><comments>https://news.ycombinator.com/item?id=47567363</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=47567363</guid></item><item><title><![CDATA[New comment by pshirshov in "Show HN: BreezePDF – Free, in-browser PDF editor"]]></title><description><![CDATA[
<p>What exactly? Agents do that for me, if I want to do something manually - I have, for example, Bento.</p>
]]></description><pubDate>Sun, 29 Mar 2026 21:09:47 +0000</pubDate><link>https://news.ycombinator.com/item?id=47567350</link><dc:creator>pshirshov</dc:creator><comments>https://news.ycombinator.com/item?id=47567350</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=47567350</guid></item><item><title><![CDATA[New comment by pshirshov in "Show HN: BreezePDF – Free, in-browser PDF editor"]]></title><description><![CDATA[
<p>Well, all the agents, including free and even local ones could do this for less money and without AGPL violations.<p>Just tell them what you need to change/merge and they literally do it just fine. Or they could write reusable python/whatever scripts for you.<p>These days $12/month for a vibe-coded PDF editor running locally is a robbery.<p>Also, let me quote:<p>>  BentoPDF (12.3k stars): <a href="https://github.com/alam00000/bentopdf" rel="nofollow">https://github.com/alam00000/bentopdf</a><p>> PDFCraft (3.6k stars): <a href="https://github.com/PDFCraftTool/pdfcraft" rel="nofollow">https://github.com/PDFCraftTool/pdfcraft</a><p>> PDFLince (31 stars): <a href="https://github.com/GSiesto/pdflince" rel="nofollow">https://github.com/GSiesto/pdflince</a></p>
]]></description><pubDate>Sun, 29 Mar 2026 21:04:04 +0000</pubDate><link>https://news.ycombinator.com/item?id=47567292</link><dc:creator>pshirshov</dc:creator><comments>https://news.ycombinator.com/item?id=47567292</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=47567292</guid></item><item><title><![CDATA[New comment by pshirshov in "Velxio 2.0 – Emulate Arduino, ESP32, and Raspberry Pi 3 in the Browser"]]></title><description><![CDATA[
<p>From what I can see, I cannot give it an ESP32-C6 firmware binary to execute, I have to compile my source code (C only) and the source has to use Arduino APIs. Under the hood it's still QEMU.</p>
]]></description><pubDate>Sat, 28 Mar 2026 20:37:32 +0000</pubDate><link>https://news.ycombinator.com/item?id=47557947</link><dc:creator>pshirshov</dc:creator><comments>https://news.ycombinator.com/item?id=47557947</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=47557947</guid></item><item><title><![CDATA[C++ standard libray with automated memory management]]></title><description><![CDATA[
<p>Article URL: <a href="https://github.com/pg83/std">https://github.com/pg83/std</a></p>
<p>Comments URL: <a href="https://news.ycombinator.com/item?id=47555414">https://news.ycombinator.com/item?id=47555414</a></p>
<p>Points: 4</p>
<p># Comments: 0</p>
]]></description><pubDate>Sat, 28 Mar 2026 15:22:32 +0000</pubDate><link>https://github.com/pg83/std</link><dc:creator>pshirshov</dc:creator><comments>https://news.ycombinator.com/item?id=47555414</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=47555414</guid></item><item><title><![CDATA[New comment by pshirshov in "We rewrote JSONata with AI in a day, saved $500k/year"]]></title><description><![CDATA[
<p>Engineers are afraid of writing custom parsers and interpreters.</p>
]]></description><pubDate>Fri, 27 Mar 2026 10:17:49 +0000</pubDate><link>https://news.ycombinator.com/item?id=47540873</link><dc:creator>pshirshov</dc:creator><comments>https://news.ycombinator.com/item?id=47540873</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=47540873</guid></item><item><title><![CDATA[New comment by pshirshov in "90% of Claude-linked output going to GitHub repos w <2 stars"]]></title><description><![CDATA[
<p>Yeah, but all these internal and not so internal tools I baked with it are great - they solve my own problems - and without LLMs I would never have a chance to implement even 20% of that.</p>
]]></description><pubDate>Thu, 26 Mar 2026 00:25:50 +0000</pubDate><link>https://news.ycombinator.com/item?id=47525228</link><dc:creator>pshirshov</dc:creator><comments>https://news.ycombinator.com/item?id=47525228</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=47525228</guid></item><item><title><![CDATA[New comment by pshirshov in "Western carmakers' retreat from electric risks dooming them to irrelevance"]]></title><description><![CDATA[
<p>Lane keeping stops working for a minute, so the car suddenly feels like a dumb one.</p>
]]></description><pubDate>Sun, 22 Mar 2026 10:53:21 +0000</pubDate><link>https://news.ycombinator.com/item?id=47476221</link><dc:creator>pshirshov</dc:creator><comments>https://news.ycombinator.com/item?id=47476221</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=47476221</guid></item><item><title><![CDATA[New comment by pshirshov in "Western carmakers' retreat from electric risks dooming them to irrelevance"]]></title><description><![CDATA[
<p>> poor public charging infrastructure in Germany<p>Dunno, had a trip through it last year, there are more than enough chargers. Some of them were literally free.<p>I have 70kWh battery though. Also, I paid much less than 40k for my chinese SUV. The software is buggy though, a random reboot on motorway doesn't feel nice.</p>
]]></description><pubDate>Sat, 21 Mar 2026 15:48:15 +0000</pubDate><link>https://news.ycombinator.com/item?id=47468082</link><dc:creator>pshirshov</dc:creator><comments>https://news.ycombinator.com/item?id=47468082</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=47468082</guid></item><item><title><![CDATA[New comment by pshirshov in "RX – a new random-access JSON alternative"]]></title><description><![CDATA[
<p>Looks similar to <a href="https://github.com/7mind/sick" rel="nofollow">https://github.com/7mind/sick</a></p>
]]></description><pubDate>Thu, 19 Mar 2026 09:35:05 +0000</pubDate><link>https://news.ycombinator.com/item?id=47436848</link><dc:creator>pshirshov</dc:creator><comments>https://news.ycombinator.com/item?id=47436848</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=47436848</guid></item><item><title><![CDATA[New comment by pshirshov in "More than 135 open hardware devices flashable with your own firmware"]]></title><description><![CDATA[
<p>> more than 135<p>136?</p>
]]></description><pubDate>Wed, 18 Mar 2026 11:19:27 +0000</pubDate><link>https://news.ycombinator.com/item?id=47424222</link><dc:creator>pshirshov</dc:creator><comments>https://news.ycombinator.com/item?id=47424222</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=47424222</guid></item><item><title><![CDATA[New comment by pshirshov in "The Isolation Trap: Erlang"]]></title><description><![CDATA[
<p>I believe it's more correct to reference circular calls as "livelocks", not "deadlocks" - something is happening but the whole computation cannot progress.<p>For the rest - pure untyped actors come with a lot of downsides and provoke engineers to make systems unnecessarily distributed (with all the consistency and timeout issues). There aren't that many problems which can be mapped well directly to actors. I personally find async runtimes with typed front-ends (e.g. Cats/ZIO in Scala, async in Rust, etc) much more robust and much less error-prone.</p>
]]></description><pubDate>Sat, 14 Mar 2026 13:16:18 +0000</pubDate><link>https://news.ycombinator.com/item?id=47376384</link><dc:creator>pshirshov</dc:creator><comments>https://news.ycombinator.com/item?id=47376384</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=47376384</guid></item><item><title><![CDATA[New comment by pshirshov in "Please Do Not A/B Test My Workflow"]]></title><description><![CDATA[
<p>Where the value you extract out of the model is orders of magnitude higher than the price of 2..6 hours of your time.</p>
]]></description><pubDate>Sat, 14 Mar 2026 12:57:50 +0000</pubDate><link>https://news.ycombinator.com/item?id=47376218</link><dc:creator>pshirshov</dc:creator><comments>https://news.ycombinator.com/item?id=47376218</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=47376218</guid></item><item><title><![CDATA[New comment by pshirshov in "Claude Code conducts A/B tests on core features"]]></title><description><![CDATA[
<p>> I pay $200/month for Claude Code<p>Which is still very cheap. There are other options, local Qwen 3.5 35b + claude code cli is, in my opinion, comparable in quality with Sonnet 4..4.5 - and without a/b tests!</p>
]]></description><pubDate>Sat, 14 Mar 2026 12:46:27 +0000</pubDate><link>https://news.ycombinator.com/item?id=47376122</link><dc:creator>pshirshov</dc:creator><comments>https://news.ycombinator.com/item?id=47376122</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=47376122</guid></item><item><title><![CDATA[New comment by pshirshov in "Kotlin creator's new language: talk to LLMs in specs, not English"]]></title><description><![CDATA[
<p>From what I was able to understand during the interview there, it's not actually a language, more like an orchestrator + pinning of individual generated chunks.<p>The demo I've briefly seen was very very far from being impressive.<p>Got rejected, perhaps for some excessive scepticism/overly sharp questions.<p>My scepticism remains - so far it looks like an orchestrator to me and does not add enough formalism to actually call it a language.<p>I think that the idea of more formal approach to assisted coding is viable (think: you define data structures and interfaces but don't write function bodies, they are generated, pinned and covered by tests automatically, LLMs can even write TLA+/formal proofs), but I'm kinda sceptical about this particular thing. I think it can be made viable but I have a strong feeling that it won't be hard to reproduce that - I was able to bake something similar in a day with Claude.</p>
]]></description><pubDate>Thu, 12 Mar 2026 16:47:21 +0000</pubDate><link>https://news.ycombinator.com/item?id=47353661</link><dc:creator>pshirshov</dc:creator><comments>https://news.ycombinator.com/item?id=47353661</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=47353661</guid></item><item><title><![CDATA[New comment by pshirshov in "Claude Code Remote Control"]]></title><description><![CDATA[
<p>In this case you are locking your workflow to the vendor's solution.</p>
]]></description><pubDate>Wed, 25 Feb 2026 14:34:58 +0000</pubDate><link>https://news.ycombinator.com/item?id=47152044</link><dc:creator>pshirshov</dc:creator><comments>https://news.ycombinator.com/item?id=47152044</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=47152044</guid></item></channel></rss>