Posts by matt@toot.cafe
(DIR) Post #Au4itqNCMO1mVBOw6K by matt@toot.cafe
2025-05-14T05:18:48Z
0 likes, 1 repeats
My #RustWeek conference talk about #AccessKit is coming up in about 3 hours, at 10:25 CET. That's 8:25 UTC, or 3:25 AM US Central time (my usual time zone). It will be live-streamed here: https://rustweek.org/live/wednesday/ My talk is in the main track.
(DIR) Post #AuoYz3dwzYSXZSt7Wy by matt@toot.cafe
2025-06-04T15:25:09Z
0 likes, 0 repeats
@baldur At least in the posts I've seen, the authors seem to be seeing evidence that the tools can in fact deliver, at least in some contexts, particularly coding. So I've concluded that a decision not to use these tools has to be based purely on principle. I hope I'm somehow wrong about that. I hope there are practical reasons too.
(DIR) Post #AuoYz5l77tNU8GFryS by matt@toot.cafe
2025-06-04T15:31:13Z
0 likes, 0 repeats
@baldur Have you seen the case of the Cloudflare Workers OAuth provider library (https://github.com/cloudflare/workers-oauth-provider). Particularly see the README section titled "Written using Claude". Kenton says he was a skeptic, intended to prove his skepticism with this experiment, and was surprised. He documented the process, including prompts. He's certain he developed the library faster this way than he would have otherwise.
(DIR) Post #AuoYz6xYf6vTr95K0e by matt@toot.cafe
2025-06-04T15:32:57Z
0 likes, 0 repeats
@baldur That project was featured on Hacker News the other day. To me, it's the most convincing of the recent burst of posts. But I still believe it's not _right_ to use these tools.
(DIR) Post #AxSbEyrSahUQ6FBNDc by matt@toot.cafe
2025-08-23T10:05:21Z
0 likes, 0 repeats
Does anyone have experience with any of the recent RP2040 or RP2350-based boards that have DVI/HDMI output, a USB host port or ports, SD card slot, and audio? These boards are often used to run retro computer emulators. I'm particularly interested in the quality of the audio output, and whether anyone has been able to use that feature from Rust. Pinging @raph in particular, who wrote this HN comment: https://news.ycombinator.com/item?id=44883045
(DIR) Post #AxSbF3OZjWM6A6hsOG by matt@toot.cafe
2025-08-23T10:27:35Z
0 likes, 0 repeats
Damn, the most promising-looking board (https://www.adafruit.com/product/6200) is out of stock. The Olimex RP2350pc (https://www.olimex.com/Products/RaspberryPi/PICO/RP2350pc/open-source-hardware) is in stock, but it mentions PWM stereo audio generation, and my first Raspberry Pi project over 10 years ago taught me to hate the original Pi's PWM-based audio. Though that page _also_ mentions an audio codec, so I don't know if the PWM audio is all there is or just one of two options.
(DIR) Post #AxSbF3STV1TUMCWzT6 by matt@toot.cafe
2025-08-23T10:07:17Z
0 likes, 0 repeats
Yes, I'm itching to do an embedded Rust project on bare metal. Even Linux is too complex for my tastes these days. I think the growing complexity of modern software, as epitomized by the generative AI craze, is making me want to run the other way.
(DIR) Post #AxhmApO1tBYsQI2Ox6 by matt@toot.cafe
2025-08-30T18:34:39Z
1 likes, 1 repeats
Congratulations to @YaLTeR and @DataTriny for getting screen reader accessibility implemented in the latest release of the niri Wayland compositor, using #AccessKit. https://github.com/YaLTeR/niri/releases/tag/v25.08
(DIR) Post #AxieZhLMKDgKF563e4 by matt@toot.cafe
2025-08-31T05:26:36Z
0 likes, 0 repeats
Is there any ARM64 SoC that can boot to Linux without either requiring a DRAM controller initialization/training blob or punting the whole boot process to a second processor running wholly closed firmware? That second one rules out Raspberry Pi (for personal projects). I'd grudgingly accept ARM Trusted Firmware persistently running in the background if the SoC's branch/fork of that is fully open-source. But I still don't like how much code has to run before you get to the OS kernel.
(DIR) Post #AxsJf9JNc3PkyVWoNs by matt@toot.cafe
2025-09-04T21:27:05Z
1 likes, 0 repeats
@ariadne > that someone is trying to improve on this by integrating screen reader hints and other accessibility features into wayland itself is an improvement over X11.Hi. I think you're talking about my project. It's been on hold for a year now; the last status update was: https://blogs.gnome.org/a11y/2024/06/18/update-on-newton-the-wayland-native-accessibility-project/What I like about my approach is that accessibility tree updates are serialized, unlike any existing platform accessibility API I know of... so they could efficiently be pushed over a network.
(DIR) Post #AxzIihWO8cnA60Si3s by matt@toot.cafe
2025-09-07T23:59:23Z
0 likes, 0 repeats
For anyone who thinks Wayland compositors are being overly restrictive by not allowing applications to set absolute positions for their windows, as I did, there are good reasons for this restriction. https://canonical-mir.readthedocs-hosted.com/stable/explanation/window-positions-under-wayland/
(DIR) Post #AxzIilU3PFIkNvOtyi by matt@toot.cafe
2025-09-08T00:02:25Z
0 likes, 0 repeats
Yes, this is a problem for some applications. But such applications can still run under Xwayland, and that will likely remain the case for a long time. So I think the Wayland community's approach of carefully, deliberately addressing specific use cases through protocol extensions is correct, though it's annoying to application developers in the short term.
(DIR) Post #Ay1RHoNJpt1jRAOPYW by matt@toot.cafe
2025-09-09T00:36:09Z
1 likes, 0 repeats
Thinking about the attack on certain popular npm packages that made news today, it occurs to me that my favorite language's package manager (Rust's Cargo) is far too much like npm for comfort. Proponents and creators of languages with no standard language package manager (especially C and the various C alternatives other than Rust) are posting their takes.
(DIR) Post #Ay1RHsrx7vuLNKkvrM by matt@toot.cafe
2025-09-09T00:38:15Z
0 likes, 0 repeats
But I still like Rust and don't want to give it up; C and the other C alternatives don't have nearly as good a story when it comes to memory safety. So that leaves me with no good answer to the kind of problem that has been brought to light (again) by today's attack, except, I guess, to just be more selective about what dependencies we add, and be a little more willing to write straightforward though possibly tedious code ourselves rather than adding a dependency that could become a liability.
(DIR) Post #AzAyLNIsTI2KzTG2YC by matt@toot.cafe
2025-10-13T13:01:48Z
1 likes, 0 repeats
Has anyone ever done a systematic study of what causes non-web applications, especially desktop apps, to start up slowly? I have my hunches, but they're just that, not anything scientific.
(DIR) Post #AzAyLSHHycjkPzYLGi by matt@toot.cafe
2025-10-13T13:03:05Z
0 likes, 0 repeats
It continues to be a problem, even in these days of NVMe SSDs. It seems to especially be a problem with some products that have a long history and thus, probably, a lot of technical debt.
(DIR) Post #AzAyLXQgp6EAOazQ7E by matt@toot.cafe
2025-10-13T13:05:43Z
0 likes, 0 repeats
And the obvious follow-up question is: how does one design an application for fast startup and maintain that over the long term? If I remember correctly, this is an explicit goal of Chromium, or at least used to be.
(DIR) Post #AzdmhyCcUiPn4AKXse by matt@toot.cafe
2025-10-27T16:51:42Z
0 likes, 0 repeats
@baldur How about Ghost? Can't it do email newsletters?
(DIR) Post #Azs4aJnWq1xL78XBIW by matt@toot.cafe
2025-11-03T13:18:16Z
1 likes, 2 repeats
I recently saw a toot saying that Linux on the desktop will take off and go mainstream once the terminal is hidden away under advanced settings (actually, twenty years after that). I had an immediate emotional reaction to that. This was my response: https://toot.cafe/@matt/115485907341571353I feel like future generations need to be able to easily discover programming, almost stumble into it, as I could on my family's first computer, an Apple IIGS. So the idea of burying the terminal just seems wrong.
(DIR) Post #Azs4aPIuMn1mBuFoP2 by matt@toot.cafe
2025-11-03T13:30:50Z
1 likes, 0 repeats
Putting a terminal front and center is a statement that you, the user, will want or even need to do things with your computer that we, the developers, didn't necessarily anticipate and design prefab, polished user interfaces for. Burying the terminal is a statement that we developers know best, and you users are a lower class that need to be protected from the things that we think you can't handle.