Sign in

june

@paperdoll.computer
103 followers 474 following 207 posts

you have to live

PostsRepliesMedia
june @paperdoll.computer · 10/10/2026
i suppose can either take a small performance hit, keep it in the same compilation unit, or the language could support cross compilation unit inlining, (this could work via the source, it doesn’t have to be link time), regardless they’re no worse off for having the tools
000
june @paperdoll.computer · 10/10/2026
that is a good point, although imo map and other fp primitives are important enough to special case if the compiler can’t optimise it. my previous lang compiled map to an inline for loop, but prepending it into the unit would do too, (this could be against the simplicity goals of some projects)
100
june @paperdoll.computer · 09/10/2026
gc aside i don't really believe efficiency is an excuse:
A screenshot of godbolt compiler explorer, the C code shows a simple static map function using a for loop, a static square function, and a test function that calls map square on an array. The assembly is very simple, the calls to square and map have been inlined.

The C code reads:
#include <stddef.h>

static void map(const int *in, int *out, size_t n, int (*f)(int)) {
    for (size_t i = 0; i < n; i++)
        out[i] = f(in[i]);
}

static int square(int x) { return x * x; }

void test(const int *in, int *out, size_t n) {
    map(in, out, 5, square);
}
The assembly reads:
"test":
        xor     eax, eax
.L2:
        mov     edx, DWORD PTR [rdi+rax*4]
        imul    edx, edx
        mov     DWORD PTR [rsi+rax*4], edx
        add     rax, 1
        cmp     rax, 5
        jne     .L2
        ret
120
june @paperdoll.computer · 05/10/2026
genuinely thank you so much
040
june @paperdoll.computer · 21/09/2026
thanks!
010
june @paperdoll.computer · 21/09/2026
how does generating its own tasks work? is there anything you can link me to on that? (i don’t know much about ML sorry)
111
june @paperdoll.computer · 21/09/2026
what kind of tasks are you evaluating it on?
110
june @paperdoll.computer · 21/09/2026
how did this go?
110
june @paperdoll.computer · 16/09/2026
(assuming they had a settings GUI, better mouse support and a smoother learning curve) i hate that so many open source gui apps have strayed from a unix file and pipe orientated approach to a more windows/mac like all in one experience with no automation possible
011
june @paperdoll.computer · 16/09/2026
yeah that’s true that they’re copying ios/macos, and people hate them for it! it does does seem that people do want win95 for some reason. i personally think the desktop metaphor was a mistake, and that tiling WMs would actually be simpler and more productive for everyone
110
june @paperdoll.computer · 16/09/2026
i don’t really see how this is true when the largest linux desktop (gnome) has strayed far from the old windows ux. no start menu, no desktop icons. i don’t know the numbers but i suspect that a majority of linux users are using either gnome or a tiling wm, both of which are far from windows
110
june @paperdoll.computer · 23/07/2026
i really want there to be .arpa tld handles, unfortunately they explicitly disallow this: atproto.com/specs/handle
atproto.com
Handle - AT Protocol
A specification for human-friendly account identifiers.
030
june @paperdoll.computer · 07/07/2026
which firmware is this on?
000
june @paperdoll.computer · 14/06/2026
how do you think it could have gone down?
000
june @paperdoll.computer · 03/05/2026
regarding a, the reason this is coming up is because of the democratic primary, in which there are more than two options. all this preemptive hypothetical vote shaming only signals that evil-lite is good enough, you can and should ask for more BEFORE it actually comes down to two options
100
june @paperdoll.computer · 03/05/2026
or you could be choosing the better of many acceptables? this argument is assuming that in all cases the options are a) limited to two, and b) both evil. another thing is that in the system i assume you’re talking about, many people’s votes are already useless.
120
june @paperdoll.computer · 01/05/2026
unsure of how this guy ever remembered his email address, it’s got a dot and whatever this thing “@“ is! so confusing!
010
june @paperdoll.computer · 28/04/2026
oh god i hate that you’re right about this
030
june @paperdoll.computer · 28/04/2026
“pay me or i’ll make branding for your app to make it look like ai slop” sounds like coercion, just don’t use images, you don’t have to pay anyone you couldn’t do this for a big brand, just because small projects don’t have branding guidelines and lawyers doesn’t mean anything goes
030
june @paperdoll.computer · 04/04/2026
you’ve reinvented DNS but worse and ActivtyPub, walking back several of the innovations of atproto
010
june @paperdoll.computer · 03/04/2026
considering that code is “ephemeral” to him that’s not a bad assumption T_T
030
june @paperdoll.computer · 01/04/2026
genuine question: does attie do anything that couldn’t be done with an advanced search/filter type UI?
100
june @paperdoll.computer · 31/03/2026
voicelink looks like macos only, what do you recommend for linux?
100
june @paperdoll.computer · 30/03/2026
maybe some text above your own like list (or the setting you mention) warning you would be enough, like they do for the public profile toggle
020
june @paperdoll.computer · 30/03/2026
this is a good point, however the difference is that users already know that blocks are public, it's impossible for them not to be. with likes i think that not showing them in the UI can give a false sense of privacy that could catch many people out. there needs to be better communication regardless
110
june @paperdoll.computer · 30/03/2026
anyone can already see all your likes via third party tools, not showing them on the profile is a false sense of privacy. it’s necessary that they are public in order to count likes and display them on posts and in notifications. i’ve linked my likes to give an example of what the data looks like
pdsls.dev
at://paperdoll.computer/app.bsky.feed.like
List of app.bsky.feed.like records from paperdoll.computer
130
june @paperdoll.computer · 30/03/2026
I wouldn't call that "well described" personally, it's a bit of a strange grouping and heading to put spam under.
000
june @paperdoll.computer · 29/03/2026
they exist, but imo the main problem is that the report flow doesn't have a well described option for bots or spam. you can't report an account like you describe to a labler accurately. there are also no notifications for verdict, so the effort of reporting is completely unacknowledged (maybe good?)
000
june @paperdoll.computer · 29/03/2026
same, i can't wait to get my homelab and setup tranquil
tangled.org
tranquil.farm/tranquil-pds
Our Personal Data Server from scratch!
010
june @paperdoll.computer · 29/03/2026
why is this a good idea when they want to encourage third parties?
100
june @paperdoll.computer · 29/03/2026
the people are begging, literally begging for private data and real 2FA, instead we are presented with ai powered nonsense that competes with the AT ecosystem :) making a feed (the old way) is one of the easiest ways to learn about the protocol and serve your community, why discourage it?
020
june @paperdoll.computer · 28/03/2026
launching an ai powered feed generator, competing with companies with Graze that have popped up in the ecosystem is such a strange decision. haven’t they been telling people that 1. they’re all about the protocol not the product, and 2. building on atproto is a good business decision
150
june @paperdoll.computer · 28/03/2026
wow this is an amazing feature
010
june @paperdoll.computer · 22/03/2026
joke aside i think multisig for did:plc updates (transferring space owne) is possible and could be useful
010
june @paperdoll.computer · 15/03/2026
didn’t they nerf the featureset? i think part of the problem it that usb-c docks aren’t common enough, if every library and uni let you plug in your own device to use their screen, keyboard, mouse and network instead of needing to use their computer it would change things
120
Reposted by june
ivy sly @ivysly.com · 25/12/2025
mathematicians be like group theory set theory and category theory are all different things
3354
june @paperdoll.computer · 15/12/2025
is that a bad thing?
120
june @paperdoll.computer · 28/11/2025
sorry if i’m missing sarcasm but what do you mean by “nobody pirates popular things, they're bulletproof”, huge amounts of people torrent anime and streaming service shows, and pirate AAA games rather then pay for 5 streaming services or shell out $90 a title
010
june @paperdoll.computer · 27/11/2025
oauth scopes aren’t finished yet, so the developer has to ask for those permissions for now.
010
june @paperdoll.computer · 05/11/2025
would be great if people stopped spreading atproto misinformation (this is being worked on and there are ways to add private data)
010
june @paperdoll.computer · 03/11/2025
thank you this is so nice to have
000
june @paperdoll.computer · 29/10/2025
this could be enforced at the appview/client level right? removing all !no-discover users from all feeds that the user does not follow (yes 3rd party client/appview's could ignore this). it seems better then leaving it up to feed authors at least
100
june @paperdoll.computer · 28/10/2025
really cool that non of the people getting mad in this thread replied to your actual example
000
june @paperdoll.computer · 20/10/2025
“pretty sure” yeah okay
100
june @paperdoll.computer · 20/10/2025
@gork.it is this true
110
june @paperdoll.computer · 20/10/2025
do you live on a yacht?
110
june @paperdoll.computer · 16/10/2025
i hate to tell you that bookmarks are stored in the appview (unless something changed recently)
030
june @paperdoll.computer · 15/10/2025
can anyone set the trans theme?
100
june @paperdoll.computer · 14/10/2025
the bigger losers are the people that brought into any of the thousands of schemes, its better to be overly skeptical and there is no immediate benefit to the majority of people, with a lot to lose
100