<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: debazel</title><link>https://news.ycombinator.com/user?id=debazel</link><description>Hacker News RSS</description><docs>https://hnrss.org/</docs><generator>hnrss v2.1.1</generator><lastBuildDate>Wed, 23 Sep 2026 02:11:01 +0000</lastBuildDate><atom:link href="https://hnrss.org/user?id=debazel" rel="self" type="application/rss+xml"></atom:link><item><title><![CDATA[New comment by debazel in "AX – Google’s Open Agentic Orchestrator"]]></title><description><![CDATA[
<p>Yes, but what I'm saying is that it is much easier to put a limit on the read_file tool or the http tool rather than sandboxing your whole environment.<p>You only really need to sandbox when you provide access to tools that are almost impossible to filter correctly, such as a bash tool or a tool for arbitrary code execution.</p>
]]></description><pubDate>Mon, 21 Sep 2026 09:27:07 +0000</pubDate><link>https://news.ycombinator.com/item?id=49785011</link><dc:creator>debazel</dc:creator><comments>https://news.ycombinator.com/item?id=49785011</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49785011</guid></item><item><title><![CDATA[New comment by debazel in "AX – Google’s Open Agentic Orchestrator"]]></title><description><![CDATA[
<p>It is really easy to restrict and contain an AI agent as long as you don't give it access to a terminal. If you only give it tools to read files and access the internet, then it is much easier to just restrict the tools themselves rather than setting up a whole isolated sandbox.</p>
]]></description><pubDate>Mon, 21 Sep 2026 07:07:18 +0000</pubDate><link>https://news.ycombinator.com/item?id=49783967</link><dc:creator>debazel</dc:creator><comments>https://news.ycombinator.com/item?id=49783967</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49783967</guid></item><item><title><![CDATA[New comment by debazel in "GPT-6 Astra"]]></title><description><![CDATA[
<p>ARC's harness is just straight up broken. No serious harness removes reasoning context between each step. Not only does this significantly lower performance over all reasoning LLMs, but it also increase cost as you destroy the cache on every turn. Tossing the oldest entry when context fills up instead of using compaction is equally bad with the same issues.</p>
]]></description><pubDate>Thu, 03 Sep 2026 20:31:50 +0000</pubDate><link>https://news.ycombinator.com/item?id=49556473</link><dc:creator>debazel</dc:creator><comments>https://news.ycombinator.com/item?id=49556473</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49556473</guid></item><item><title><![CDATA[New comment by debazel in "Nitter and XCancel receive cease and desist notices"]]></title><description><![CDATA[
<p>Not possible, says phone number is invalid.</p>
]]></description><pubDate>Wed, 26 Aug 2026 01:18:38 +0000</pubDate><link>https://news.ycombinator.com/item?id=49443003</link><dc:creator>debazel</dc:creator><comments>https://news.ycombinator.com/item?id=49443003</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49443003</guid></item><item><title><![CDATA[New comment by debazel in "We must not grant AI agents legal personhood"]]></title><description><![CDATA[
<p>Is it really anymore dangerous than having minimum wage workers serving as fall guys for approving prompts they have no understanding over?</p>
]]></description><pubDate>Sun, 23 Aug 2026 18:42:43 +0000</pubDate><link>https://news.ycombinator.com/item?id=49411289</link><dc:creator>debazel</dc:creator><comments>https://news.ycombinator.com/item?id=49411289</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49411289</guid></item><item><title><![CDATA[New comment by debazel in "Codex in ChatGPT desktop app for Linux is now in preview"]]></title><description><![CDATA[
<p>It did it during the installation process.</p>
]]></description><pubDate>Thu, 13 Aug 2026 12:53:29 +0000</pubDate><link>https://news.ycombinator.com/item?id=49285199</link><dc:creator>debazel</dc:creator><comments>https://news.ycombinator.com/item?id=49285199</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49285199</guid></item><item><title><![CDATA[New comment by debazel in "Codex in ChatGPT desktop app for Linux is now in preview"]]></title><description><![CDATA[
<p>I never sent any prompts to it, it did all of this during the installation. Here is one related GitHub issue I found for it <a href="https://github.com/openai/codex/issues/12343" rel="nofollow">https://github.com/openai/codex/issues/12343</a> Looks like they don't intend to change this.</p>
]]></description><pubDate>Thu, 13 Aug 2026 12:52:27 +0000</pubDate><link>https://news.ycombinator.com/item?id=49285189</link><dc:creator>debazel</dc:creator><comments>https://news.ycombinator.com/item?id=49285189</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49285189</guid></item><item><title><![CDATA[New comment by debazel in "Codex in ChatGPT desktop app for Linux is now in preview"]]></title><description><![CDATA[
<p>And that's not a joke, I made the mistake of installing this on my Windows machine just to test it out quickly last week.<p>It created 2 new users and then assigned new NTFS permissions for every single file under my user directory to them. This of course wrecked havoc, ssh refused to work, several applications refusing to start and a ton of permission errors. It did this without even a warning in the background and it also does not undo any of it when you uninstall it.<p>It took around 3 hours to fix it by updating ~10 million NTFS permissions for every single file under my user directory.</p>
]]></description><pubDate>Thu, 13 Aug 2026 10:54:31 +0000</pubDate><link>https://news.ycombinator.com/item?id=49284113</link><dc:creator>debazel</dc:creator><comments>https://news.ycombinator.com/item?id=49284113</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49284113</guid></item><item><title><![CDATA[New comment by debazel in "Facebook ads are so hard to block that uBlock Origin stopped filtering them"]]></title><description><![CDATA[
<p>We're not that far off. But it will not just be for AI, but all software. Mobile OSes has already normalized the expectation that you don't control what can run on your own devices.</p>
]]></description><pubDate>Wed, 12 Aug 2026 16:41:19 +0000</pubDate><link>https://news.ycombinator.com/item?id=49275170</link><dc:creator>debazel</dc:creator><comments>https://news.ycombinator.com/item?id=49275170</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49275170</guid></item><item><title><![CDATA[New comment by debazel in "Why tiny JPEGs look different in Chrome"]]></title><description><![CDATA[
<p>I'm far from an image rendering expert myself, but I don't think there is a single root cause here. I don't doubt the partial decoding would affect the final result just as the scaling algorithm itself, so in the end it's going to be a combination of it all.<p>The reason I believe the scaling algorithm is contributing more in your particular example, is because it looks almost exactly like a classic B/W comparison between a sharp ringing heavy algorithm and a more blurry one.<p>For reference here is a quick comparison between Lanczos/Bilinear I threw together: <a href="https://twinlens.app/compare?share=40973d84a174" rel="nofollow">https://twinlens.app/compare?share=40973d84a174</a><p>It would be interesting to see how your example would look if you used colors instead, where the ringing artifacts become a lot more obvious. Like this one: <a href="https://twinlens.app/compare?share=8fb890a5820e" rel="nofollow">https://twinlens.app/compare?share=8fb890a5820e</a></p>
]]></description><pubDate>Wed, 12 Aug 2026 16:20:02 +0000</pubDate><link>https://news.ycombinator.com/item?id=49274848</link><dc:creator>debazel</dc:creator><comments>https://news.ycombinator.com/item?id=49274848</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49274848</guid></item><item><title><![CDATA[New comment by debazel in "Why tiny JPEGs look different in Chrome"]]></title><description><![CDATA[
<p>Chrome and Firefox uses two different scaling algorithms that is probably contributing a lot more to this difference. Chrome is more blurry in general while Firefox is sharper but has slightly more ringing artifacts. Personally I prefer the Firefox version.</p>
]]></description><pubDate>Wed, 12 Aug 2026 14:50:22 +0000</pubDate><link>https://news.ycombinator.com/item?id=49273283</link><dc:creator>debazel</dc:creator><comments>https://news.ycombinator.com/item?id=49273283</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49273283</guid></item><item><title><![CDATA[New comment by debazel in "Show HN: Mcptoon – Token-efficient MCP CLI client"]]></title><description><![CDATA[
<p>Why is it replacing true/false with T/F? true/false is already 1 token in all tokenizer I've seen. Even worse is replacing null with ∅. ∅ is a special unicode symbol that takes up 2 tokens compared to the 1 token for null...</p>
]]></description><pubDate>Tue, 11 Aug 2026 06:56:11 +0000</pubDate><link>https://news.ycombinator.com/item?id=49254312</link><dc:creator>debazel</dc:creator><comments>https://news.ycombinator.com/item?id=49254312</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49254312</guid></item><item><title><![CDATA[New comment by debazel in "Windows 11's built-in Weather app wastes more than 1 GB of RAM"]]></title><description><![CDATA[
<p>On Windows over allocating virtual memory (commit charge) is still bad. Unlike on Linux, once the total commit charge hits your physical RAM limit + SWAP limit, Windows will start crashing your applications and refuse to allocate more virtual memory, even if that memory has not been touched yet.<p>This is also why it is very important to have plenty of SWAP space on Windows, even if you have 64 GiB+ of memory. Because applications love to over allocate commit charge.</p>
]]></description><pubDate>Sun, 09 Aug 2026 19:30:24 +0000</pubDate><link>https://news.ycombinator.com/item?id=49234841</link><dc:creator>debazel</dc:creator><comments>https://news.ycombinator.com/item?id=49234841</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49234841</guid></item><item><title><![CDATA[New comment by debazel in "DeepSeek V4 Flash 0731"]]></title><description><![CDATA[
<p>I've been using this DeepSeek model the whole day today after building with 5.6 Luna extensively over the last week and I would disagree, at least for Rust + OpenGL.<p>DeepSeek just spend almost 2 hours trying to figure out why terrain textures were not working. It tried everything over and over again, it even had reference code for meshes on how to setup the rendering with materials, and it could just not do it.<p>I finally gave up and gave it to GPT-5.6 Luna instead, and figure out in a single prompt after 20 seconds, that the terrain mesh was being initialized with None in the material slot.<p>Other tasks it has managed to figure out at least, but it is significantly slower than GPT-5.6 Luna and it requires a lot more iterations.<p>(Both were set to high reasoning)</p>
]]></description><pubDate>Sat, 08 Aug 2026 10:06:43 +0000</pubDate><link>https://news.ycombinator.com/item?id=49220347</link><dc:creator>debazel</dc:creator><comments>https://news.ycombinator.com/item?id=49220347</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49220347</guid></item><item><title><![CDATA[New comment by debazel in "Kill The Cookie Banner"]]></title><description><![CDATA[
<p>Users complained that we were non-compliant with EU regulations due to the missing cookie banner.<p>Because of the legal wording of these support tickets, the support team would often raise them to dev or legal as well, and since this is a non-revenue driving part of our business (the reason we don't track in the first place) we decided to just put up a cookie banner explaining why we don't need a cookie banner.</p>
]]></description><pubDate>Tue, 28 Jul 2026 15:43:29 +0000</pubDate><link>https://news.ycombinator.com/item?id=49085609</link><dc:creator>debazel</dc:creator><comments>https://news.ycombinator.com/item?id=49085609</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49085609</guid></item><item><title><![CDATA[New comment by debazel in "Kill The Cookie Banner"]]></title><description><![CDATA[
<p>One of our sites that don't use any kind of tracking or cookies other then a site preference cookie, got so many complaints that we had to put up a cookie banner that simply says "we do not use any tracking cookies"...</p>
]]></description><pubDate>Sun, 26 Jul 2026 19:12:41 +0000</pubDate><link>https://news.ycombinator.com/item?id=49061363</link><dc:creator>debazel</dc:creator><comments>https://news.ycombinator.com/item?id=49061363</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49061363</guid></item><item><title><![CDATA[New comment by debazel in "Passkeys were invented by engineers with zero understanding of consumer brain"]]></title><description><![CDATA[
<p>Only to other proprietary systems within the FIDO Alliance that you don't own or control.<p>Additionally passkeys allow services to detect and ban specific password managers, so have fun when the only approved managers that works consistently across all services are Google/Apple/Microsoft. There is already a list of "bad" clients here <a href="https://passkeys.dev/docs/reference/known-issues/" rel="nofollow">https://passkeys.dev/docs/reference/known-issues/</a></p>
]]></description><pubDate>Wed, 22 Jul 2026 15:37:49 +0000</pubDate><link>https://news.ycombinator.com/item?id=49008518</link><dc:creator>debazel</dc:creator><comments>https://news.ycombinator.com/item?id=49008518</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=49008518</guid></item><item><title><![CDATA[New comment by debazel in "Deno Desktop"]]></title><description><![CDATA[
<p>ironically the only group of users I've found that actually care about native UI, are other developers and Mac purist.</p>
]]></description><pubDate>Mon, 22 Jun 2026 07:19:28 +0000</pubDate><link>https://news.ycombinator.com/item?id=48626851</link><dc:creator>debazel</dc:creator><comments>https://news.ycombinator.com/item?id=48626851</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=48626851</guid></item><item><title><![CDATA[New comment by debazel in "Developers don't understand CORS (2019)"]]></title><description><![CDATA[
<p>Is there a reason this has to happen client side with extra pre-flight requests? Taking your example, why couldn't Facebook's server just check the origin header and then reject all request from unapproved origins server side instead?</p>
]]></description><pubDate>Sun, 21 Jun 2026 10:47:39 +0000</pubDate><link>https://news.ycombinator.com/item?id=48617646</link><dc:creator>debazel</dc:creator><comments>https://news.ycombinator.com/item?id=48617646</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=48617646</guid></item><item><title><![CDATA[New comment by debazel in "Stop Using JWTs"]]></title><description><![CDATA[
<p>A revocation list can often be lazily replicated and doesn't require the complexity of distributed synchronization.<p>If you store sessions then you need to ensure the session has been replicated to every node before you can return it to the user. For revocation lists it is often acceptable that the token is still valid for a short while at some nodes while it is being replicated.<p>A revocation list is also not considered highly sensitive data, which would be another complexity layer when working with distributed data.</p>
]]></description><pubDate>Wed, 17 Jun 2026 09:02:01 +0000</pubDate><link>https://news.ycombinator.com/item?id=48567685</link><dc:creator>debazel</dc:creator><comments>https://news.ycombinator.com/item?id=48567685</comments><guid isPermaLink="false">https://news.ycombinator.com/item?id=48567685</guid></item></channel></rss>