thenickdude

Backported security patches don't bump the version string.

I have a beanie with your name on it somewhere! It was free swag from the adhesives company.

>The biggest trap is the hallucinated citation

The "biggest problem" being the one thing that is trivial to verify against concrete databases is a bit convenient don't you think?

I think it's more likely that it makes mistakes evenly but the one thing that you are able to check with certainty is the only place you discover the errors.

LXD uses dnsmasq to provide DHCP and DNS for containers I think? Viable container escape?

I delete the third party libraries and replace them with my own clean checkouts.

If you ever produce and share a signed link for e.g. S3, this link contains your access key ID in it. Turns out you can just slice and decode your Account ID out of that access key, it's in there in base32:

https://medium.com/@TalBeerySec/a-short-note-on-aws-key-id-f...

Deep in the abyss, the dark lord stirs

I was trying out an IDE plugin for local LLM integration, but it would just make totally insane edits instead of what I asked for.

It turned out that the LLM-runner's default setting for handling queries that were too long was to silently delete the middle of the query. So the LLM saw the beginning of my ask, and the tail end of the code context, and nothing else.

I've run Japanese windows software under Wine, and with no configuration it was a sea of crashy mojibake. With the right locale configured, it worked fine.

At the bottom of each issue is a huge amount of whitespace, then they've stuffed it with a bunch of @'s to notify users.

https://i.imgur.com/LRotRlS.png

The link goes to a typo-squatted domain using an "l" instead of an "i".

It's all in aid of some streetsweeper being able to add "contributor to X, Y, Z projects!" to their GitHub résumé. Before LLMs were a thing I also received worthless spelling-incorrection pull requests with the same aim.

This means you should still have the email from legal@, right? In that case you can solve the mystery of how they managed to pass DMARC by sharing the headers from it.

This was not Google's decision, it was forced on them by the EU under their Digital Services Act:

https://9to5mac.com/2024/10/17/developers-address-phone-numb...

The same thing was applied to the Apple store at the same time.

We have that already, we call it "Stack Overflow"

Interestingly when I asked GPT-4o (at least that's what it said it was):

>According to the instructions, which UnicodeCIDFont fonts should be used when generating PDFs?

It replies:

>When generating PDFs using reportlab for East Asian languages, you must use specific UnicodeCIDFont fonts depending on the language. According to the instructions, use the following:

>Korean: HeiseiMin-W3 or HeiseiKakuGo-W5 or HYSMyeongJo-Medium

>Simplified Chinese: STSong-Light

>Traditional Chinese: MSung-Light

>These fonts must be registered using pdfmetrics.registerFont(UnicodeCIDFont(font_name)) and applied to all text elements in the PDF when outputting those languages.

This list also has the Japanese fonts merged with the Korean list.

https://chatgpt.com/share/6895a4e6-03dc-8002-99d6-e18cb4b3d8...

They've released their report now, 10 further certificates were mis-issued:

https://bugzilla.mozilla.org/show_bug.cgi?id=1961406

The free tier now supports connecting to local AI models running on LM Studio or Ollama, but it still doesn't actually function without an internet connection.

If you block access to the internet or to their AI API servers [1], it refuses to start a new chat invocation. If you block access halfway through a conversation, the conversation continues just fine, so there's no technical barrier to them actually running offline, they just don't allow it.

Their settings page also says that they can't even guarantee that they implemented the offline toggle properly, a flag that should be the easiest thing in the world to enforce:

>Prevents most remote calls, prioritizing local models. Despite these safeguards, rare instances of cloud usage may still occur.

So you can't even block access to the very servers that they say their faulty offline toggle would leak data to.

[1] https://www.jetbrains.com/help/ai-assistant/disable-ai-assis...

1x 3090 (350W power limit) already makes it feel like I'm running a fan heater under my desk, 5x would be nuts.

Japan has a finders' fee of 5-20%:

https://www.police.pref.ehime.jp/foreigner/otoshimono/2.pdf

Except 100 yen stores which are actually 110 yen stores.

You can buy crypto on PayPal?? Our New Zealand version doesn't mention it either.

Only two total MIPI ports (DSI/CSI, for displays and cameras respectively) compared to 4 ports on the CM4.

Sampling Bias: https://en.wikipedia.org/wiki/Sampling_bias

A sandbox service identifies it as Lumma Stealer, so it'll at least steal all your passwords, cookies and cryptocurrency, and then anything else after that is fair game too:

https://socradar.io/malware-analysis-lummac2-stealer/

You can also use the Apple AAC encoder using ffmpeg (on a Mac only), with the argument "-c:a aac_at". Handy as it allows you to use it for encoding the audio track of videos, in addition to its support for pure audio container formats:

https://trac.ffmpeg.org/wiki/Encode/AAC#aac_at

Same here on my issue tracker, seems to be GitHub wide.

https://i.imgur.com/jJljJp8.png

They reply with the message to new issues, which makes their notification email look like a direct solution to the submitter's issue, so they'll probably catch out a lot of people...

Surely they would at least be protected by Database Rights in the EU (not the US):

>The TRIPS Agreement requires that copyright protection extends to databases and other compilations if they constitute intellectual creation by virtue of the selection or arrangement of their contents, even if some or all of the contents do not themselves constitute materials protected by copyright

https://en.wikipedia.org/wiki/Database_right

New Zealand