Hacker Newsnew | past | comments | ask | show | jobs | submit | ValdikSS's commentslogin

That's right, it's easier to setup such MiTM using an intermediate server, because only getting the private key of the certificate won't get you the user's traffic due to PFS.

You either need to disable PFS on the server, or export TLS master keys for each session in some way, or MiTM.


Bad at translation, at least to Russian. Very fast though, about 2x faster than Gemma 4 e2b on my CPU.

On GitHub, if anyone makes an MR which changes the name of existing GitHub Actions, it will be instantly renamed. The web interface will show redefined name, even if the MR comes from new contributor (and yet unmerged of course).

You're biased. AI was trained on a well-written professional texts which used these phrases and speech patterns, it was very common before the AI-generated texts. These speech patterns especially common for Russian speakers.


I can't understand why people defend improper typography. If you're writing a proper, professional-looking blog post, they think you now should use double-minus -- instead of em-dash to make it look non-AI like, only for that reason?

In Russia, we have many typography keyboards/addons, because, well, it historically looked very silly to use double-minus or "-quotes instead of «»-quotes.

I've no idea how some countries got their typography standardized on the PCs and have it from the very beginning (Germany with their quotes for example), but the other countries need to setup external software and configuration. Apparently, US also didn't got their "third level" keyboard as a standard.


It's a great concept, but you haven't open-sourced the previous code, as the license requires, and you're yet again apologizing in this project as well, without any code.

Pretty sure you have my code in both projects. I contribute first and foremost to make printers and scanners to work reliably, but also keeping in mind the idea that I could at least try to apply legal actions for companies which violate the license rules one day, as a CUPS/SANE/printer/scanner drivers contributor.

Printer companies generally don't like that: https://xcancel.com/ValdikSS/status/1745898408693371125#m

Cool project though! Hope you can publish the source one day so we can all benefit from it in the future!


If you think I’ve done something wrong according to the licences involved here, please do clarify. I had understood that open-sourcing the Linux stuff (as branches of a fork of v86, linked from the /credits page) met all relevant legal obligations, which I absolutely intend to do.

More broadly, it’s unusual for me not to make everything open, and I do feel bad/conflicted about it. But, unusually, I feel like I have identified a possible route to monetising this, and I think open-sourcing all of it risks making that harder.


Sorry, it's me who needs a reading comprehension lessons. I've read back in printervention website and now again that you didn't open the code that you HAVE to. Because you're apologizing for that, I assumed that you're breaking the license, twice.

After rereading both of your websites again, I should say you've nothing wrong! It's sleepy me who accused you for nothing, sorry.

Linux printing and scanning stack is held on 5 enthusiasts basically, and is quite buggy. Any contributions welcome.

If you want to further improve your project, make it small and fast, you can compile printer filters (most of which work on cups-raster data) with emscripten. This way you don't need to use CUPS, Linux, and x86 emulation. You'll need to write some shims for CUPS libppd functions which many filters use (some don't), and either parse PPD files or convert them into another representation.

Most filters (drivers) are quite simple pipes from stdin to stdout, sometimes they don't use cups functions at all, receiving all the data directly from raster header. Some filters, such as gutenprint, are more complex and use their own backends, but even in this case it's not a hard task: libusb has emscripten WebUSB backend.


OT: But in a way kind of good to know.

Ages ago I got a Canon A3 printer. I've never been sure if it worked properly, as I was never sure if the colours are right.

Next time I unpack it I'll have to try and find the place the 5 enthusiasts hang out - the cups mailing list ?


If monetization is at odds with open-source, why wouldn't potential customers just wouldn't go to VueScan, as someone posted? I was recently looking at scanners, and saw some brands directly advertise Linux support through this... which means you now have to pay subscription each year to access the expensive hardware you bought.

Thankfully the Avision FB5100 states native Linux support (AFAIK, this is the only flatbed A3 scanner that does), so I'm certainly going to buy this one. I know implementing device support for companies that don't make any effort is hard and thankless, but then we need to divest/invest in the right companies and solutions.


Any airprint/mopria certified devices don't need drivers to work on Linux, Windows, Android or macOS.

https://mfi.apple.com/account/airprint-search

https://mopria.org/certified-products


My recent experience shows that eSCL is way behind in terms of functionality. If I want lossless scanning from by Brother scanner, I need the proprietary drivers.


My monetization idea doesn't involve charging users, and it's more on the printing side (but most of the source is shared with scanning).


If you just install CUPS in a virtual machine (emulated in wasm on the web) what patches do you need to share?


See above, that's my mistake.


The same you have to share if you don't use a virtual machine, this isn't hard.


hope we see that code soon


Linux kernel contribution policy required sending patches under real name, but that policy have been lifted about 2 years ago. Now they allow pseudonym contributions.

https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/lin...

https://github.com/cncf/foundation/blob/659fd32c86dc/dco-gui...


So my mission critical infrastructure depends on a group whose bar for entry is having a proton mail account.

I bet they claimed to be protecting trans people to get that policy changed too.


It's a part of PDF, so if there's a PDF renderer which makes preview, it supports G4 and JBIG2.


That's why LLM will eventually be used only for initial interaction between the user in their language, to prepare the data to a specialized model.

Imagine face recognition to work like a text chat, where the PC gets the frame from the camera and writes in the chat: "Who's that? Here's the RGB888 image in hex: ...".


That's actually how vision language models already work, pretty much.


And there's a reason nobody uses them for face recognition

Vision language models are an incredible achievement in the generality and usability. But they pay a hefty price in fidelity and speed


Huh? The images are tokenized in the same way language is and it’s just fed into one single model. Not multiple smaller expert models.

Image gets rasterized into smaller pieces (eg 4x4 pixels) and each of those is assigned a token, similarly how text is broken up into tokens. And the whole thing is fed into a single model.


Yes I'm saying

> Imagine face recognition to work like a text chat, where the PC gets the frame from the camera and writes in the chat: "Who's that? Here's the RGB888 image in hex: ...".

that's p much how it works.


But that isn’t a specialized model like the grandparent claimed, but rather a single, multi-modal model.


Yes, the "imagine" was showcasing the opposite of a specialized model to call it a bad idea.


Do you know that MoE is a thing?


The experts in MoEs aren't specialized in any meaningful task sense. From level of what we would think as tasks MoEs are selected essentially arbitrarily per token and per block.


It’s unsupervised, yes, but “unspecialized in any meaningful task sense” is incorrect, that’s the whole point. It’s just not in the sense of “this is a legal expert, this is a software developer”.


Optimal expert separation depends on the goal and can be pretty arbitrary, for example DeepSeek v4 separates them more or less by domain if I remember correctly.


Denuvo is owned by Irdeto, a digital rights management company in a broad sense. They not only do software and hardware DRM, but also work as a watchdog for movie and music companies to claim DMCA violations for BitTorrent, among all other stuff.


Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: