moody

As someone who has spent a decent bit of time with 9front I can not recommend this work as a medium to learn about the Plan 9 operating system unfortunately. The author has also admitted to using LLMs for some of this work.

2026-06-01 17:21:06 GMT

All I wanna do today is play some Minecraft and listen to Dr. Menzes talk about PQC.

2026-05-22 01:51:50 GMT

- reply

This was not "hidden" I was pretty up front about it in irc. But feel free to point and laugh at me as you typically do.

2026-05-21 17:20:46 GMT

My comment on the recent "Acme AI agent" thread on 9fans stating that I'm done was censored (for some reason, not even remotely heated) so instead I am posting some thoughts here.
I have over the past maybe 4 years or so attempted to rebuild the bridge somewhat between 9legacy folks and 9front. This was in-part driven by iwp9 happening and generally having a good time getting to meet folks outside of the immediate 9front sphere. Part of this was driven by the fact that having two semi-hostile groups is confusing for newcomers, part of this was in hopes to reduce duplicate work, part of it was looking to allow us to do more by working together. I was not here when this fissure between these two groups started, so I thought perhaps we could bury the hatchet and everyone would be better off.
Trying to do this is messy, and I'll be the first to admit that I'm no saint, if you go looking you'll find plenty of heated posts from me to 9fans. However, slowly bit by bit, I felt like things were slowly improving. Folks were airing grievances and for once it felt like there was some actual listening and improvements. One part of this was the 9front git force push to get rid of some jokes that haven't aged well, the final list of which was done in conjunction with the p9f for something everyone agrees with. The hopes of doing that was that with this clean up we could finally get some recognition on the main websites (p9f.org et all), however despite us doing the work of the force push nearly a year ago(which involved a risky and involved change to our update mechanism) there has been no change to any recognition.
As we'll see this becomes a recurring theme, expectations that 9front developers and maintainers should shoulder the lion-share of the work burden for doing something. Over the past couple of years we've gotten drive-by PRs made by some 9legacy folks, and when there is any sort of additional questions or requests for changes the response we get is almost one of insult. I want to be more charitable but it comes across as "how do you question MY change to your shitty fork". As soon as someone says anything the result is "well I'm just going to take my toys and go home then". The final boss of this was this vibe coded crapola PR sent to us for us to roll back a bunch of uses of our new interfaces in git (many of which were done to make the code more maintainable) to make it compile out of the box on 9legacy.
Not to mention that after being clearly stated that 9front does not accept computer generated software, no one was willing to answer the question of if they used it. No one wants to admit anything inconvenient. No one wants to communicate or solve problems together. They don't want to make 9front better, they want to use it for their own uses and make that our problem. So I'm sorry but I just don't to work with the p9f and their community if this is the result. Other 9front community members warned me this would be the outcome but I wanted to believe otherwise.
Its important to note that I dont have anything against any of these folks personally, I'd happily have a drink with anyone, at the end of the day we have something in common if we're still using Plan 9, but liking someone and working together are two different things. I also want to make a carve out for Richard Miller, who has carved his own path and continuously been working (mostly) in the open, large amount of respect for him keeping on.

2026-05-21 16:35:33 GMT

I was wondering why I was getting noise about this article I wrote from 2 years ago, turns out it was reposted to lobsters two days ago (and then cross posted to reddit). Despite my article being a reflection of how poorly the conversation around this vague "philosophy" is, and how the history is a bit more nuanced than people expect, all of the comments opt instead to just state their own ideas in vague relation to this vaguely defined philosophy and all that is created is more of the same noise I was complaining about in the first place.

2026-04-01 04:16:11 GMT

https://hj.9fs.net/moody/s/post-492137853c8e08d97813097c2b3124ca.png

2026-03-31 15:21:10 GMT

- reply

The change in question[0] was even made by an employee of Anthropic.
[0] https://github.com/tmux/tmux/pull/4744

2026-03-30 02:58:39 GMT

& knuckles
https://hj.9fs.net/moody/s/post-d9a53c48dcd0a6f71df985111cbf5cff.png

2026-03-26 05:32:13 GMT

For those wondering, yes you can indeed calculate a mega drive rom checksum from the comfort of your 9front rc window:
% { echo 'n=0'; xd -2w streetsofrage.md | grep -v '[0-9a-z]+ $' | sed '1,32d;s/^.* /n=trunc(n+0x/g;s/ /+0x/g;s/$/,16)/g'} | pc | tail -n 1
0x9409

2026-03-24 04:05:05 GMT

Part of the reason I played around with the TMSS is because of these statements about this Sonic 3D Blast 814 prototype from tcrf.net[0].
Essentially stating that the ROM contained the string "Nosegay is a word." somewhere in the high end parts of the rom, and this was done specifically as a way to bypass TMSS checks. To be clear neither this or any of the Phys-Q dev kit games actually use this value, likely just bundled in with the provided software kit. Perhaps a known "legal backup" to be deployed after a C&D.
Looking through the tmss disassembly it seems like it would work, it only checks that next 4(5) bytes at 0x100 contain either "SEGA" or "SEGA" with a leading space. However you have two leading bytes you need to stuff before 0x100, which would bleed in to a copy of the ErrorTrap interrupt handler address, which very possibly is not used at all on hardware. However you still have to deal with the TMSS register inside of the IO controller itself which must be filled with 2 values containing exactly "SEGA". You could fill this straight from the rom header data without having to embed an explicit "SEGA" constant again. So it seems like yes, NOSEGAY is a valid mega drive rom header as long as you define the header to start 2 byte early.
I was curious because the overdrive demo uses a header value of "SEGA MEGA DRIVET", so I was curious what the exact checks are.
[0] https://tcrf.net/Proto:Sonic_3D_Blast_(Genesis)/814_Prototype

2026-03-23 21:56:54 GMT

My rabbit hole for last night: https://git.9front.org/plan9front/9front/598072cc6d9e221e47dc1a5958fb0b11d074db40/commit.html
games/md now can fully make use of a standard mega drive tmss rom. Not sure I've seen other emulators do this but I'm not super well versed. Would be fun to take it as a demo challenge, just 4k of rom to do something with.
https://hj.9fs.net/moody/s/post-8b26cd2efeed72406bc96fa2661e5d98.png
https://hj.9fs.net/moody/s/post-a2b9096a198f308666ab9ac53a30ba34.png

2026-03-14 05:52:50 GMT

SEGA mapper in games/md.
https://hj.9fs.net/moody/s/post-4267117f9715a3fd3ce9a66b889ad88a.png

2026-02-04 23:20:26 GMT

Reading the latest UTC notes, it seems the C2PA (a group tasked with marking AI content with some sort of watermark, sponsored by some of the worst offenders ironically), has decided the best way to tack on "metadata" to plain text as AI is to encode arbitrary data into a string of Unicode variant selectors[0] slapped onto a Zero-Width No-Break space (U+FEFF). This is not only a relatively ineffective watermark being completely trivial to strip, but is a pretty flagrant abuse of variant selectors.
[0] https://spec.c2pa.org/specifications/specifications/2.3/specs/C2PA_Specification.html#embedding_manifests_into_unstructured_text

2026-01-31 01:01:50 GMT

I did not expect 9fans to start posting lots of AI generated code, but in hindsight I should have seen it coming.

2026-01-19 15:28:46 GMT

- reply

I'm not aware of any issues (at least that I've seen or heard about). I know vulture sometimes does rolling upgrades when a vm is restarted, did the issues coincide with that?

2026-01-15 17:43:21 GMT

- reply

Just a bit, let's see if it captures my attention enough to max this time.

2026-01-14 19:19:32 GMT

The reason why 9front work has been slow lately.
https://hj.9fs.net/moody/s/post-5e636e40861bf0540cb1ca62ce8f75fe.png

2025-12-20 04:31:20 GMT

- reply

Hellaphone lives again.

2025-09-11 19:52:54 GMT

- reply

but but but what about my USE flags? How else am I going to get my software with 2000 knobs tuned to just my liking?

2025-09-10 03:24:33 GMT

Day 1 Unicode v17.0 support on 9front for normalization, grapheme break and word break.

2025-07-19 01:41:36 GMT

Testing the beta for Unicode v17.0 for 9front, thankfully this time no changes are needed for our normalization algorithm. All lights are green, and we're prepared to launch on release day.

2025-05-01 06:15:23 GMT

I've been mulling over the idea of attempting to create a custom build of 9front for the purpose of using it as a networking appliance. A NAT router box, some p2p vpn thing, etc. Something akin to OPNsense in a way. Would people be interested in that? What would you do with it? Is this a terrible waste of time?

2025-04-10 04:57:21 GMT

note: someone will make fun of you for this
https://hj.9fs.net/moody/s/df9a512e928a36d29a8accdccf119627.png

2025-03-09 19:39:34 GMT

- reply

Fantastic.

2025-03-08 18:52:03 GMT

- reply

There was an initial pass for unicode support, to at least not barf on it, but there was some lingering issues that prevented its use in identifiers.

2025-03-08 05:45:37 GMT

9front's yacc now supports Unicode in identifiers, can your water buffalo do that?

2025-02-09 00:29:47 GMT

I may have mentioned this before but wanted to again in case people haven't seen, 9front does have a riscv64 userspace (albeit not very well tested) and what we're really missing is someone working on a kernel (even if only for qemu). I did a decent chunk of that work and would be willing to act as support for the userspace(compilers libraries and so on) portions if someone wanted to try taking on that project. There is a start to a kernel, enough to jump in to virtual memory. I've still got the kernel portions of the power9 port to work on but considering the userspace stuff is at a decent place acting as support is doable.

2025-02-06 03:20:42 GMT

[@[email protected]](https://hachyderm.io/users/robpike ) During the original work on Unicode support in Plan 9, was there any consideration for the Non-Spacing Marks, like the Diacritical attaching runes? I'm just curious if it was due to a lack of approval of those portions of the standard or if it was avoided out of the amount of technical work needed to support it given the current code.

2024-10-10 11:27:23 GMT

Happy teto day
https://hj.9fs.net/moody/s/a2d217ab5e5f66b5f2c25ce671cf46ee.jpg

2024-06-18 16:54:38 GMT

- reply

That's true, I had forgot about those proposals. Perhaps microsoft pushing these snapdragon consumer laptops might also contribute to some change. It would be interesting to see a new x86 ISA version.

2024-06-18 16:11:37 GMT

- reply

Unless someone is a nerd wanting to work on something related to an ISA (compiler, kernel and so on) I would largely say that the choice of ISA isn't much an impact, outside of perhaps software compatibility and prebuilt packages. However there are forces at play that color what the priorities are for each ISA. ARM wants power efficiency, x86 wants backwards compatibility. While I don't doubt a "no legacy and cleaned up" x86 would do quite well, I don't think we'll see it.

2024-06-15 14:51:35 GMT

Next part in the video series I'm doing on Plan 9 design, this time talking about file abstractions: https://www.youtube.com/watch?v=r8X5i6BFZiY

2024-06-15 01:11:50 GMT

https://www.theregister.com/2024/06/11/niconico_ebiten_kadokawa_cyberattack_outages/ RIP nico nico, crazy that they would get hit like this

2024-06-14 17:52:58 GMT

- reply

Yeah I finally got audio with screen sharing in firefox thanks to some nixos flake, so all the more reason to just use it from there.

2024-06-14 17:41:01 GMT

- reply

yeah using it in the browser works. Discord is running via XWayland, so it still uses the X11 method for screen sharing, which limits it to only sharing other applications running under XWayland. It sucked as it was, but I could at least stream video games to my friends.

2024-06-14 16:05:31 GMT

Screen sharing with discord on linux has been broken for like 3 months with wayland.

2024-06-13 16:02:35 GMT

- reply

https://github.com/bilbof/whoishiring-ratio

2024-06-13 15:46:14 GMT

Job market sucks.

2024-04-08 17:51:34 GMT

Thinking about how prof(1) could be better. There is a lot to dislike about the tooling as is, but the hooks the linker generates are quite powerful. I'd like a flame graph tool and the death of $profsize for starters.

2024-04-08 03:52:59 GMT

Took a look at porting tree sitter to 9front today but they do some cursed stuff with bit fields to satisfy some alignment assumption their code makes so nah not happening.

2024-04-07 04:36:46 GMT

RIP miku expo, crunchyroll strikes again.

2024-04-04 15:05:22 GMT

- reply

I think it really gets an unfairly bad wrap. The first time I think I really "got it" as a structural tool was reading the libdraw code in Plan 9. There it is used for function unwinding so every allocation has a corresponding label that will jump to the next one to slowly undo any heap allocations made by a function in the event of an error.

2024-04-03 22:18:46 GMT

Wrote an updated blog post about the current status of the Talos 2. Not too far along with the port but wanted to correct my previous post about getting it working with some help from Raptor. https://posixcafe.org/blogs/2024/04/03/0/

2024-04-03 21:34:05 GMT

- reply

Yeah give those a look, they tend to be a bit on the larger side so if you want something more smaller and focused I might be able to find something. Or if you have some ideas about what type of projects you like to work on more broadly maybe we could find something that fits.

2024-04-03 20:36:19 GMT

- reply

Yeah sure, do you have an idea about things you want to work on or would be interested in? If not there are some ideas listed https://orib.dev/tmp/student-ideas.txt and [http://p9f.org/wiki/gsoc-2024-ideas/index.html.](http://p9f.org/wiki/gsoc-2024-ideas/index.html)

2024-04-03 19:17:23 GMT

Just throwing this out there, but if anyone is new to Plan 9/9front and either don't know a project to start working on or are intimidated to get going I can always help get going. Send me an email or a DM.

2024-04-03 17:35:15 GMT

I would be very interested to see if someone could get a reproducible build of 9front going. Probably isn't _too_ hard, just need enough to get kencc compiled (and by extension yacc). Would make for a cool IWP9 paper.

2024-04-02 16:03:26 GMT

Thanks to a patch from the community the Korean input method for 9front has improved greatly. In general I've been really happy with how ktrans has turned out.

2024-04-01 19:14:21 GMT

Also just for the record I am really not some long time ppc fan or anything. My current impression from reading the ISA so far is that it's like just fine, but maybe I just don't get it yet. I'm more interested in this process of porting than I am in the power64 ISA itself. The talos 2 just happened to exist at the intersection of "is a desktop" and "need a port to run 9front".

2024-04-01 15:19:33 GMT

- reply

Just how the MMU works. The SLB and HPTAB both work pretty different compared to other other systems. POWER9 does have a radix implementation these days but I want to try the "true" ppc method.

2024-04-01 03:59:22 GMT

Memory address translation on power64 is what I would call quirky.