Conversation
Notices
-
Embed this notice
pistolero (p@fsebugoutzone.org)'s status on Tuesday, 08-Apr-2025 13:54:38 JST 
pistolero
@scathach @Suiseiseki @jeffcliff
> a standardized mechanism for exchanging structurally complex data that doesn't require writing ad-hoc parsers on both ends
A parser is just a means of transferring wire-format data into an internal data structure. You cannot have what you are suggesting unless the data structure is the same on both sides, and you cannot get that without coupling the programs so tightly that you may as well abandon pretense at this being IPC.
You don't need to exchange structurally complex data anyway. You need distributed CSP, and you already have it.- Phantasm likes this.
-
Embed this notice
翠星石 (suiseiseki@freesoftwareextremist.com)'s status on Tuesday, 08-Apr-2025 15:08:27 JST 
翠星石
@lispi314 @p @jeffcliff @scathach You should avoid having a standard system ABI, as that makes it easy to develop proprietary malware for the system.
Only a standard system API should be implemented.pistolero likes this. -
Embed this notice
LisPi (lispi314@udongein.xyz)'s status on Tuesday, 08-Apr-2025 15:08:28 JST 
LisPi
@p @scathach @jeffcliff @Suiseiseki There's no reason not to have a standard system ABI and/or an environment for hosting programs atop of that permits for abstracting those concerns.
This doesn't require strongly coupling programs. They can remain separate, and written in separate languages. All that is required is a compiler adequate for the target environment and/or some abstraction interface library to interface with said ABI or environment properly & transparently.
Whether structured messages can be passed directly or require ABI interpretation shouldn't be the programs' concern beyond that. -
Embed this notice
翠星石 (suiseiseki@freesoftwareextremist.com)'s status on Tuesday, 08-Apr-2025 15:27:33 JST
翠星石
@lispi314 @p @jeffcliff @scathach Source-available proprietary malware is quite rare, as anyone who cares to read the source finds out what malicious features the program implements.
Most of such programs are not in fact source-available and have obfuscated sections for the malware parts that are not source code.pistolero likes this. -
Embed this notice
LisPi (lispi314@udongein.xyz)'s status on Tuesday, 08-Apr-2025 15:27:34 JST
LisPi
@Suiseiseki @scathach @jeffcliff @p That (no ABI) requires a hosted language-based system in order to have adequate structured messaging capacity.
It also requires the language to have a fairly rich type-system, otherwise that will be a wasted effort and guest languages will yet again have to formulate supplementary interchange formats will all sorts of unwanted slowdowns as a result.
That being said, source availability is irrelevant to a program's nature as proprietary malware. Source-available proprietary malware is readily available.pistolero repeated this. -
Embed this notice
pistolero (p@fsebugoutzone.org)'s status on Wednesday, 09-Apr-2025 02:01:22 JST
pistolero
@lispi314 @Suiseiseki @jeffcliff @scathach
> Whether structured messages can be passed directly or require ABI interpretation shouldn't be the programs' concern beyond that.
Application programmers and systems programmers are not going to agree on this statement or its implications. -
Embed this notice
the_daikon_warfare (sicp@freesoftwareextremist.com)'s status on Wednesday, 09-Apr-2025 02:14:39 JST 
the_daikon_warfare
@lispi314 @Suiseiseki @p @jeffcliff @scathach
You don't need " a hosted language-based system" to parse lines of text or TSV or something. I'm all for S-expressions but it's not like you need a hard format to have structured data.
Also p was talking about Plan 9 which just uses the filesystem for IPC, so I don't know why you're bringing up ABIs and shit.pistolero likes this. -
Embed this notice
pistolero (p@fsebugoutzone.org)'s status on Wednesday, 09-Apr-2025 03:28:15 JST
pistolero
@sicp @lispi314 @Suiseiseki @jeffcliff @scathach
> I don't know why you're bringing up ABIs and shit.
The aforementioned salt problems, maybe. Nothing really stops a person from building out their Common Lisp monolith system, except that it's an untenable way of designing a system.
Even Unix luminaries fell victim to this blind spot: Dennis Ritchie joked that Unix had retarded operating systems research by ten years, and Linux by another twenty. Maybe that's the case, but the operating system matters less: how hard is it to
The Internet Protocol is, so far, the only really successful effort at cross-platform communication and HTTP is the most successful effort at a universal application-layer protocol for cross-platform communication. The latter is somewhat painful for everyone. It's more painful when you tack on "x86 won" and "the single-user PC won" and "Unix won" and then "Linux won" but that is all history now. (If you put them all next to each other, history contradicts itself: Unix was designed around multiple people hacking together, and now it's the operating system for these single-user desktop systems.)
Hamming is one of the few credible people, I think, that hated ALGOL. McIlroy and McCarthy were both on the committee, most of the team that created Unix loved the language, and ALGOL's influence is easy to spot. But Hamming noted--accurately--that ALGOL was designed by language designers that wanted to create a beautiful language, and FORTRAN was designed to be used by the people that were going to be using it. (He referred to this as "logical design versus psychological design".) Brian Kernighan, when asked in an interview what he thought of ML, remarked that, although he didn't want to disparage his colleagues' work, he couldn't help noticing that ML had almost exclusively been used to write ML compilers.
If you look at the scant decades we have for language and OS design, simple designs with room around the edges (i.e., "undefined behavior") always wins. Cray builds a computer, he's got a bunch of cheap transistors, he comes up with a design to make a reliable circuit from unreliable parts by stacking the transistors, and when he's done with his design, he's produced the fastest machine on earth.
Unix, you know, some people hate how the 80s went: Lisp people think Unix is ugly, GNU people seem to hate Unix and want to turn it into god-knows-what. So they assume because the lower level of the system is ugly (in their estimation) that everything from outside the tower is bad and lowbrow. The most important thing that happened in computing since we figured out how to use the computer to program itself is "CSP won". The system is no longer a single machine and CSP is the only workable way to model distributed software. People that are hung up on how Unix implements syscalls tend not to appreciate that Unix is CSP all the way down. I really like Lisp but the thing that keeps me from reaching for it is that the designers hate Unix (because "you're supposed to" and "there's no excuse for not having a" and all of that) so I/O sucks in every Common Lisp or Scheme implementation. I get it, to some extent: I had that problem with Lisp when I was a yoof. I like Plan 9 a lot more than I like Unix but I'd never get anything done if I let my own snobbery get in the way. (Like Linux, Plan 9 didn't start as the dominant paradigm and as a consequence, it had to grow a lot of facilities for talking to other systems over the network, so it's nice to use as a development environment regardless.)
CSP didn't just win, CSP is the only tenable way of building systems, especially if those systems touch a network. You draw a diagram of the system you want to build and if you architect it properly (i.e., by thinking of modules as communicating sequential processes) then it doesn't matter whether the lines between the bubbles cross the processes' memory space or even cross the machine by using the network.
(This turned into a rant that may not apply specifically to people in this thread. I am just exasperated that people keep stepping on their own dicks and trying to solve problems that do not need solving, or worse, complaining that nobody is solving those problems.)
:sun: The network is the computer. :sun:Phantasm likes this. -
Embed this notice
the_daikon_warfare (sicp@freesoftwareextremist.com)'s status on Wednesday, 09-Apr-2025 04:08:38 JST
the_daikon_warfare
@lispi314 @Suiseiseki @p @jeffcliff @scathach
Bullshit.pistolero likes this. -
Embed this notice
LisPi (lispi314@udongein.xyz)'s status on Wednesday, 09-Apr-2025 04:08:39 JST
LisPi
@sicp @scathach @jeffcliff @p @Suiseiseki > to parse lines of text or TSV or something. I'm all for S-expressions but it's not like you need a hard format to have structured data.
And promptly performance has just dropped by a magnitude or more.
Compiling various languages to Common Lisp (or some other host language) and running programs in its environment allows for safe near zero-cost IPC (through the use of object capabilities for maintaining isolation in memory) with no serialization overhead.
9p has a lot of such overhead. -
Embed this notice
Jeff "never puts away anything, especially oven mitts" Cliff, Bringer of Nightmares 🏴☠️🦝🐙 🇱🇧🧯 🇨🇦🐧 (jeffcliff@shitposter.world)'s status on Wednesday, 09-Apr-2025 04:14:46 JST
Jeff "never puts away anything, especially oven mitts" Cliff, Bringer of Nightmares 🏴☠️🦝🐙 🇱🇧🧯 🇨🇦🐧
@p @Suiseiseki @scathach
> You need distributed CSP,
maybe it's my lack of coffee so far but what's that in this context again?pistolero likes this. -
Embed this notice
pistolero (p@fsebugoutzone.org)'s status on Wednesday, 09-Apr-2025 04:15:23 JST
pistolero
@jeffcliff @Suiseiseki @scathach :tonyhoare: Communicating Sequential Processes. :tonyhoare: -
Embed this notice
Jeff "never puts away anything, especially oven mitts" Cliff, Bringer of Nightmares 🏴☠️🦝🐙 🇱🇧🧯 🇨🇦🐧 (jeffcliff@shitposter.world)'s status on Wednesday, 09-Apr-2025 04:15:56 JST
Jeff "never puts away anything, especially oven mitts" Cliff, Bringer of Nightmares 🏴☠️🦝🐙 🇱🇧🧯 🇨🇦🐧
@lispi314 @Suiseiseki @p @scathach
> That being said, source availability is irrelevant to a program's nature as proprietary malware.
irrelevant goes too far: clearly it starts to behave more like free software at the point where you have access to the source code alone. if you can have access to the source code *and* a compiler that is allowed to compile(ie no gpl3/tivoization/drm restrictions) & execute it that's even better. If you have a source code, a compiler, and can share & run your changes even if it's "proprietary" it starts to look and behave like free software in every way but in some sense for which it is not (ie perhaps the license restricts it in law, if not in practice) . If the law permits you to do all of these things and the source permits you to do all of these things then you've got about as much freedom as you can get unless it's an ai-like system perhaps where there's some special edge cases worth considering etc
but even in the case where you have the source code, a compiler to run it and a system which will execute compiled code and the functional ability to share changes - even if it's not legal to do so your community will behave something like a free software community with most the benefits of social solidarity, if an underground one.pistolero likes this. -
Embed this notice
Jeff "never puts away anything, especially oven mitts" Cliff, Bringer of Nightmares 🏴☠️🦝🐙 🇱🇧🧯 🇨🇦🐧 (jeffcliff@shitposter.world)'s status on Wednesday, 09-Apr-2025 04:16:20 JST
Jeff "never puts away anything, especially oven mitts" Cliff, Bringer of Nightmares 🏴☠️🦝🐙 🇱🇧🧯 🇨🇦🐧
@Suiseiseki @p @scathach @lispi314 i'd imagine it's not as rare as we think it is -- as 'available' kinda depends who you are and the line between 'source available proprietary software' and 'software that has never been distributed except to a single/couple clients, along with source code but and the understanding that they could distribute it but they never do' code gets blurry
ie pretty much all the software i ever worked on in my career was of this kind - the clients had access to the source code and were allowed to do 'whatever they wanted' with it which probably could have included releasing it as GPL3 if they desired. We were that much beholden to them. But they never did.pistolero likes this. -
Embed this notice
Jeff "never puts away anything, especially oven mitts" Cliff, Bringer of Nightmares 🏴☠️🦝🐙 🇱🇧🧯 🇨🇦🐧 (jeffcliff@shitposter.world)'s status on Wednesday, 09-Apr-2025 04:16:26 JST
Jeff "never puts away anything, especially oven mitts" Cliff, Bringer of Nightmares 🏴☠️🦝🐙 🇱🇧🧯 🇨🇦🐧
@sicp @Suiseiseki @p @scathach @lispi314 which part? pistolero likes this. -
Embed this notice
di0nysius the patomskyite (dsm@fsebugoutzone.org)'s status on Wednesday, 09-Apr-2025 04:20:49 JST
di0nysius the patomskyite
@lispi314 @sicp @Suiseiseki @jeffcliff @p @scathach
Is that limit the throttle?pistolero likes this. -
Embed this notice
Jeff "never puts away anything, especially oven mitts" Cliff, Bringer of Nightmares 🏴☠️🦝🐙 🇱🇧🧯 🇨🇦🐧 (jeffcliff@shitposter.world)'s status on Wednesday, 09-Apr-2025 04:37:55 JST
Jeff "never puts away anything, especially oven mitts" Cliff, Bringer of Nightmares 🏴☠️🦝🐙 🇱🇧🧯 🇨🇦🐧
@p @Suiseiseki @scathach thanks pistolero likes this. -
Embed this notice
pistolero (p@fsebugoutzone.org)'s status on Wednesday, 09-Apr-2025 04:38:04 JST
pistolero
@jeffcliff @Suiseiseki @scathach :bwk: -
Embed this notice
the_daikon_warfare (sicp@freesoftwareextremist.com)'s status on Wednesday, 09-Apr-2025 04:39:53 JST
the_daikon_warfare
@jeffcliff @Suiseiseki @p @scathach @lispi314
The part where he's talking about a hypothetical magic bullet that I've never even seen applied anywhere. Now don't get me wrong, if it's a real thing then I'd love to see it. But I'm incredulous of the claim that it would have no overhead, or even be a good tool for the job. Afaik object capabilities are mostly an access control thing, which if you're running all of your processes under the same VM sounds like a fool's errand. And I don't get how that answers network-transparent storage and IPC; those bits don't just emerge out of thin air. Feel free to enlighten me.pistolero likes this. -
Embed this notice
pistolero (p@fsebugoutzone.org)'s status on Wednesday, 09-Apr-2025 05:45:35 JST
pistolero
@sicp @Suiseiseki @jeffcliff @lispi314 @scathach
> In a perfect world, we'd have a choice between Plan 9 and Lisp machines, and they'd probably be used to talking to each other.
We've got half of those. I think you could assemble a plausible Lisp workstation if you build out the Lisp environment on top of the rest. That is, you do it like OpenStep/GNUStep, or like Arcan.
I think the whole dataflow programming thing the Arcan guy is doing with Pipeworld, that's one of the first actually new HCI things that I've seen in a while. Watch the guy's demo, it's great: https://arcan-fe.com/2021/04/12/introducing-pipeworld/ . Anyway, he's using a more or less conventional Linux system for that, but there's nothing forcing it to behave that way: you could do similar with Lisp, but you suggest it and it's even odds you hear "We already have emacs, which is already perfect, please let us stagnate in peace." There's also Mezzano, I think that's still going (though it stalled a little a few years ago).
I might be in the minority, but I thought https://io10.dev/ was *delightful*. If it was less browsery (every system I have, something on there conflicts: one machine won't run the JS due to CSP, the keybindings don't work on another machine, file import/export is broken on another, etc.) I think I might use something like that a lot. It's basically like a very simplified acme with a little more focus on interactivity.
> People forget that GNU was a compromise.
Yes, I was being glib. At the higher level, there was a lot of cross-pollination.
> evangelism is a waste of time
I think so, yeah. Or, like, there's "Here's some cool shit this thing can do" and I like that, and then there's "this thing good, that thing bad". This was a good read, I wish more people would read it: https://www.perl.com/pub/2000/12/advocacy.html/
> reverence for the book is stretched to its absurd extreme by making claims about "achieving satori" by the mere act of reading it.
I think there are systems where the light bulb explodes and you are not the same after.
This happened when I learned how Lisp works, Forth, Unix, and it took way longer for C but I think once you really *get* C (something that people that use phrases like "glorified PDP assembler" can't do until they put that idea down and meditate on the pipe() syscall) then it's hard to take a lot of languages seriously.
And I think, like, people joke about it but it is a thing, a thing happens in your brain.Phantasm likes this. -
Embed this notice
the_daikon_warfare (sicp@freesoftwareextremist.com)'s status on Wednesday, 09-Apr-2025 05:45:36 JST
the_daikon_warfare
@p @Suiseiseki @jeffcliff @scathach @lispi314
In a perfect world, we'd have a choice between Plan 9 and Lisp machines, and they'd probably be used to talking to each other. They were both designed for individual hackers working together in a lab over a local network.
> Lisp people think Unix is ugly, GNU people seem to hate Unix and want to turn it into god-knows-what. So they assume because the lower level of the system is ugly (in their estimation) that everything from outside the tower is bad and lowbrow.
People forget that GNU was a compromise. :ignutius: was a Lisp hacker through and through, he lived and breathed that shit for years. He didn't think Unix was particularly good, but it was important enough at the time to replicate for his system. His attempt at bringing a bit of the Lisp machine workflow to Unixen was Emacs. His initial plans for morphing the Unix design into something else were much broader, but by the time Linux won out that plan kind of went astray. The Hurd actually had some neat ideas.
> I really like Lisp but the thing that keeps me from reaching for it is that the designers hate Unix (because "you're supposed to" and "there's no excuse for not having a" and all of that)
Sometimes I feel like a bit of an outsider among other Lisp hackers because I don't completely pull the wool over my own eyes. I like Common Lisp and Emacs 'cause they're the right tools for me, but preaching evangelism is a waste of time that just makes people hate you. It's like those old memes about SICP where the reverence for the book is stretched to its absurd extreme by making claims about "achieving satori" by the mere act of reading it. But this is apparently an actual part of Lisp culture, and they take it seriously. We can still learn stuff from our contemporaries; let's not get caught up in decades long culture war bullshit. I like Plan 9. It's a good design. When are we getting 9P over Chaosnet? -
Embed this notice
pistolero (p@fsebugoutzone.org)'s status on Wednesday, 09-Apr-2025 06:53:43 JST
pistolero
@lispi314 @Suiseiseki @jeffcliff @scathach @sicp
> I think everyone acknowledges that Emacs sucks for the overwhelming majority of graphics-related tasks.
It was a joke, not intended to be about emacs but about the attitude that makes language zealots insist "the final word has been written on $x". I grab a notepad out of my back pocket and scribble a reminder onto it and an emacs guy scoffs: "You wouldn't need to use a notepad if you had org-mode".
But I wasn't talking about graphics. Doing something like pipeworld (ignore the GUI part) or io10 is actually plausible in emacs: it is a different way of interacting with the computer. Think Jupyter notebooks, even something like spreadsheets. These are all plausible without a bitmap display. -
Embed this notice
LisPi (lispi314@udongein.xyz)'s status on Wednesday, 09-Apr-2025 06:53:44 JST
LisPi
@p @scathach @jeffcliff @sicp @Suiseiseki > but you suggest it and it's even odds you hear "We already have emacs, which is already perfect, please let us stagnate in peace."
I think everyone acknowledges that Emacs sucks for the overwhelming majority of graphics-related tasks.
It does considerably worse than Interlisp does at it. -
Embed this notice
the_daikon_warfare (sicp@freesoftwareextremist.com)'s status on Wednesday, 09-Apr-2025 12:46:59 JST
the_daikon_warfare
@lispi314 @Suiseiseki @p @jeffcliff @scathach
> With a language providing no direct memory access semantics, it is impossible to wrongly access resources without either a broken implementation (for which formal verification methodology is available) or broken hardware (this is a more complicated issue).
Sure, but the language's semantics shouldn't matter. A process shouldn't have to give away the resources you don't need, but whatever you get your hands on you should be able to do with what you please. If you're just sending blunt data it's not even a problem.
> Re-implementing that with near zero-cost would require RDMA and trust for all involved nodes (may be impossible in a lot of cases)
Sort of what I thought. Not that I believe in security anyway.
> This means the implementation can encapsulate resources in opaque structures which cannot be interfered with.
Sounds proprietary. :absolutely_proprietary:pistolero likes this. -
Embed this notice
LisPi (lispi314@udongein.xyz)'s status on Wednesday, 09-Apr-2025 12:47:00 JST
LisPi
@sicp @scathach @jeffcliff @p @Suiseiseki With a language providing no direct memory access semantics, it is impossible to wrongly access resources without either a broken implementation (for which formal verification methodology is available) or broken hardware (this is a more complicated issue).
This means the implementation can encapsulate resources in opaque structures which cannot be interfered with.
Implementations/compilers can provide APIs for handling cases where such memory access is required, namely hardware support.
> And I don't get how that answers network-transparent storage and IPC; those bits don't just emerge out of thin air.
Neither Erlang nor Goblins have that as zero-cost, they have considerable overhead even if one were to run the messaging for them over RDMA (which requires considerable node trust).
It may be impossible (with near-zero cost) without complete node & network handler program trust. -
Embed this notice
the_daikon_warfare (sicp@freesoftwareextremist.com)'s status on Wednesday, 09-Apr-2025 15:18:08 JST
the_daikon_warfare
@p @Suiseiseki @jeffcliff @scathach @lispi314
> I think the whole dataflow programming thing the Arcan guy is doing with Pipeworld, that's one of the first actually new HCI things that I've seen in a while
This is cool. Kinda like Oberon, huh? I'll have to look into it more.
> I might be in the minority, but I thought https://io10.dev/ was *delightful*. I think I might use something like that a lot. It's basically like a very simplified acme with a little more focus on interactivity.
Heh, I can look at this and say "sure, why not?". I couldn't really get into acme, the workflow was kind of weird for me, nothing wrong with it though. I like the mouse-centric workflow, which Emacs kind of sucks at. The Lisp machine actually did make good use of the mouse, so augmented it with my own replication of that.
> And I think, like, people joke about it but it is a thing, a thing happens in your brain.
Not denying that. But there's a line between "getting it" and actually finding it useful.
A lot of the lore around Lisp machines is so overblown because it's been gatekept by boomers that want to perpetuate the legend. If you actually emulate one and sit down with it for a while, it's not that special. It's cool, but it's cool the same way every other clunky bespoke workstation from the 80's is cool. The people who are actually in it for the historical/technical interest make up a small modicum of everyone who mouths off about them.pistolero likes this. -
Embed this notice
pistolero (p@fsebugoutzone.org)'s status on Wednesday, 09-Apr-2025 15:26:49 JST
pistolero
@sicp @Suiseiseki @jeffcliff @lispi314 @scathach
> This is cool. Kinda like Oberon, huh? I'll have to look into it more.
Yeah, it's fun. The way it works, bits of it are the author's idiosyncrasies bleeding through but the parts that are smooth are insanely smooth.
> I couldn't really get into acme, the workflow was kind of weird for me,
If you think of the whole thing like a frozen pipe, then acme clicks. (I have a thought or two percolating for a long time, maybe I will get a lazy Saturday to prototype it.)
> The Lisp machine actually did make good use of the mouse, so augmented it with my own replication of that.
Do go on.
> A lot of the lore around Lisp machines is so overblown because it's been gatekept by boomers that want to perpetuate the legend.
Yeah, and defensive too. Peter Suk was talking about bytecode VMs on the Ruby mailing list a long time back and I think he basically nailed it:
> I don't want to argue that Smalltalk was the first widespread commericial implementation of X, because some Lisp-er will inevitably tell me that it was done in 1970-something, and further come back to me with photos from an archaeological dig where paleo-lispers were doing semi Aspect-Oriented things with "around" methods and CDR was implemented by bashing the end off a rock with a large teak club. -
Embed this notice
LisPi (lispi314@udongein.xyz)'s status on Wednesday, 09-Apr-2025 15:28:15 JST
LisPi
@sicp @scathach @jeffcliff @p @Suiseiseki > Sounds proprietary. :absolutely_proprietary:
It's comparable to a lambda in Emacs Lisp (to a more limited extent, Elisp gets weird about the lexical vs dynamic situation and there are a bunch of caveats) or Common Lisp.
One can put access to some resources inside of one, and from within the language there is simply no way for anything receiving that lambda to crack it open and retrieve that resource (at least without messing with FFI and starting to parse the runtime's memory).
> A process shouldn't have to give away the resources you don't need, but whatever you get your hands on you should be able to do with what you please. If you're just sending blunt data it's not even a problem.
Consider C as the host language. It permits any function to dig into any memory it feels like. There is no simple way to have separate programs share memory without also enabling ambient authority of that sort.
And so an additional runtime capable of enforcing sane isolation properties becomes necessary.
That said, Common Lisp as standardized isn't sufficient for a single runtime with arbitrary program threads. Nesting environments without arbitrary access to others (via various dynamic/special elements) would be necessary (concepts written about in SICL, first-class global environments).pistolero likes this. -
Embed this notice
pistolero (p@fsebugoutzone.org)'s status on Wednesday, 09-Apr-2025 15:32:24 JST
pistolero
@sicp @Suiseiseki @jeffcliff @lispi314 @scathach
> You'd be surprised what you can get away with in Elisp.
Fun. -
Embed this notice
the_daikon_warfare (sicp@freesoftwareextremist.com)'s status on Wednesday, 09-Apr-2025 15:32:25 JST
the_daikon_warfare
@p @Suiseiseki @jeffcliff @scathach @lispi314
> These are all plausible without a bitmap display.
You'd be surprised what you can get away with in Elisp.pistolero likes this. -
Embed this notice
the_daikon_warfare (sicp@freesoftwareextremist.com)'s status on Wednesday, 09-Apr-2025 15:32:46 JST
the_daikon_warfare
@lispi314 @Suiseiseki @p @jeffcliff @scathach
> It's comparable to a lambda in Emacs Lisp or Common Lisp.
So it's like a closure, got it. I get bogged down in the academically opaque stuff. I'm a hacker, not a CS grad student.
> And so an additional runtime capable of enforcing sane isolation properties becomes necessary.
Unless you're doing pure message passing it sounds like you'd need either a runtime or some hairy control scheme in order to get this to work.
> Nesting environments without arbitrary access to others (via various dynamic/special elements) would be necessary
That's the way it's always been done as far as I know.
https://genode.org/index
Have you checked this out? Seems like it might be up your alley.pistolero likes this. -
Embed this notice
LisPi (lispi314@udongein.xyz)'s status on Wednesday, 09-Apr-2025 15:39:37 JST
LisPi
@sicp @scathach @jeffcliff @p @Suiseiseki I have been keeping an eye on Genode, though I am disappointed by the contribution agreement.
Working on it (rather than with it) would consequently require forking it.
> Unless you're doing pure message passing it sounds like you'd need either a runtime or some hairy control scheme in order to get this to work.
If the system is a Lisp runtime with support for first-class global environments, that one runtime is enough.
Otherwise one gets back to process-based separation using the kernel and messaging overhead goes back up.pistolero likes this. -
Embed this notice
pistolero (p@fsebugoutzone.org)'s status on Wednesday, 09-Apr-2025 15:47:05 JST
pistolero
@lispi314 @sicp @Suiseiseki @jeffcliff @scathach
> I have been keeping an eye on Genode, though I am disappointed by the contribution agreement.
The rights assignment? The scoldy "as you are aware that there are no such obligations in the GPL either" bit made me bristle but the "We can use your contributions in any of our future products" bit is kind of a dealbreaker. -
Embed this notice
RedTechEngineer (redtechengineer@fedi.lowpassfilter.link)'s status on Wednesday, 09-Apr-2025 15:51:55 JST 
RedTechEngineer
@p @Suiseiseki @sicp @jeffcliff @scathach @lispi314
fun)))))pistolero likes this. -
Embed this notice
the_daikon_warfare (sicp@freesoftwareextremist.com)'s status on Wednesday, 09-Apr-2025 15:58:43 JST
the_daikon_warfare
@p @Suiseiseki @jeffcliff @scathach @lispi314
> Do go on.
It's the menu I showed off down in the thread. I wrote it cause while Emacs does have context menus, they're tied to the toolkit they're built with and it's buggy and they suck. So I set out to (to see if it's even possible to) write a drop-in replacement in Elisp, with ergonomics I actually like and based off the style of the Zmacs menus.
Here's the source if you wanna take a look.pistolero likes this. -
Embed this notice
pistolero (p@fsebugoutzone.org)'s status on Wednesday, 09-Apr-2025 16:00:59 JST
pistolero
@sicp @Suiseiseki @jeffcliff @lispi314 @scathach Ah, yeah, it looks nice, it appears to work fine, except I think the gtk popups would horrify me.
> Here's the source if you wanna take a look.
I looked briefly, but elisp and vimscript and whatnot read pretty similar to me: a lot of editor internals, a lot of context. -
Embed this notice
the_daikon_warfare (sicp@freesoftwareextremist.com)'s status on Wednesday, 09-Apr-2025 16:12:33 JST
the_daikon_warfare
@p @Suiseiseki @jeffcliff @scathach @lispi314
Yeah thing with GTK is that it's especially shit in this context. I put a workaround in it but you're better off just using the Lucid toolkit. (You can also use the Emacs daemon without crashing the fucking X server that way -- decades old bug that GTK devs refuse to fix)pistolero likes this. -
Embed this notice
pistolero (p@fsebugoutzone.org)'s status on Wednesday, 09-Apr-2025 16:13:39 JST
pistolero
@sicp @Suiseiseki @jeffcliff @lispi314 @scathach
> decades old bug that GTK devs refuse to fix
jwz worked on Xemacs and had this to say, maybe you have seen it: https://www.jwz.org/doc/cadt.htmlPhantasm likes this. -
Embed this notice
pistolero (p@fsebugoutzone.org)'s status on Wednesday, 09-Apr-2025 20:03:18 JST
pistolero
@lispi314 @Suiseiseki @jeffcliff @scathach @sicp
> It's not even the rights assignment all that much (though it is distasteful at best),
I don't mind it if it's rights assigned to the project so that the project doesn't have to do that, but the rights assignment isn't to the project, it's this project any any other project that they might decide to release "in the future under an open-source license", and then combined with the next paragraph's grant of rights to "relicense, sublicense", it is kinda nasty. I wouldn't assign them copyright, I'd just fork.
> mandatory doxing (which they incidentally require for the corposcum behavior).
Is that how it works in Germany? (I know they have some quirks in their copyright law, like no public domain.) Or just this company's policy? -
Embed this notice
LisPi (lispi314@udongein.xyz)'s status on Wednesday, 09-Apr-2025 20:03:19 JST
LisPi
@p @scathach @jeffcliff @sicp @Suiseiseki It's not even the rights assignment all that much (though it is distasteful at best), it sucks in that aspect, but the hard no-deal is the mandatory doxing (which they incidentally require for the corposcum behavior). -
Embed this notice
pistolero (p@fsebugoutzone.org)'s status on Wednesday, 09-Apr-2025 20:06:12 JST
pistolero
@gentoobro @lispi314 @Suiseiseki @sicp @jeffcliff @scathach Oh, no, they really do want your name, address, phone number, etc.: https://genode.org/community/gca.pdf .
> "LisPi" is legally just as good as your legal name, since what matters is the intent to sign the contract.
I don't know how contract law works in Germany, but it's not like so much about weaseling out as it is that, if they want $x in order to merge your patches, you give that to them or you fork the project.
gca.pdf -
Embed this notice
gentoobro (gentoobro@shitpost.cloud)'s status on Wednesday, 09-Apr-2025 20:06:13 JST 
gentoobro
Which doesn't even matter from a legal perspective. "LisPi" is legally just as good as your legal name, since what matters is the intent to sign the contract. It's just like how illiterate people would put an "X" or how you can't weasel out of a contract by misspelling your name when you sign it.
-
Embed this notice
翠星石 (suiseiseki@freesoftwareextremist.com)'s status on Wednesday, 09-Apr-2025 20:43:18 JST
翠星石
@jeffcliff @p @scathach @lispi314 >software that has never been distributed except to a single/couple clients, along with source code
That is free software in the trivial sense, as all of its users have freedom (the clients paid for it to be written, therefore they are the copyright holder unless there is some agreement as to who the copyright goes to).
The freedom to share software also includes the freedom to choose not to share the software at all.pistolero likes this. -
Embed this notice
pistolero (p@fsebugoutzone.org)'s status on Friday, 11-Apr-2025 14:20:17 JST
pistolero
@gentoobro @Suiseiseki @sicp @jeffcliff @scathach @lispi314
> Yes, they want those things, but they don't have a true legal basis to need them.
Yeah, I know, but if they want that, then you can give them that or they are free to ignore you. -
Embed this notice
gentoobro (gentoobro@shitpost.cloud)'s status on Friday, 11-Apr-2025 14:20:18 JST
gentoobro
Yes, they want those things, but they don't have a true legal basis to need them. They probably want them out of typical hyper-greedy corpo take-and-control-everything-we-can nonsense, just in case they want to send you spam or lawsuits later on.
All GNU social JP content and data are available under the