But there’s still no way to get consistent results cross-platform without a web view, right? I guess Tauri isn’t even consistent because it relies on the system’s browser and they do differ to some degree.
My friend, inconsistency is the entire point. You now have incredible power to write truly native apps that fit in with the platform and follow all the standards. The bottleneck is now you, or whoever owns each target platform and understands what native apps feel like.
So I’m supposed to maintain a native app for each platform and try to be sure they’re aligned in every way except visuals? Sounds like a recipe for bad UX if anybody uses more than one of them.
On the contrary: using a web view is the worst possible UX for your users. It's slow, uses a ton of memory, and doesn't integrate well with the desktop. The only reason people ever did it was laziness, not because it produced a better product.
I'd argue that being slow and using a ton of memory has more to do with (very) poorly optimized websites rather than blaming browsers as a whole. Sure, they'll never be as efficient as native, but people put a lot of work into optimizing browsers and most websites are never optimized.
Memory use alone is an interesting case for comparison because websites tend to have a lot more images/videos loaded than other applications. All of those PNGs and JPEGs need to be decompressed before you can paint them to the screen which makes them much larger.
> I'd argue that being slow and using a ton of memory has more to do with (very) poorly optimized websites rather than blaming browsers as a whole.
A web browser is a huge platform. Audio, 2D and 3D graphics, file handling, network requests,... A lot of code while your application needs are tiny. That is wasteful and a security risk.
That's really only argument against shipping a browser with your application. For ordinary websites (or WebViews) that's not really a concern because it is shared and automatically updated.
Do people really want this? IME the worst, slowest apps on a given OS are native.
I remember a time when you could feel that apps were built on a consistent set of standards on a platform, and that meant they were fast, and it was weird when an app had its own wacky GUI. But today, native apps are neither consistent across eachother, nor are they performant.
WebView2 is native on Windows and the practical default, but who wants more apps like Microsoft Teams? Music.app is native on macOS and slower than music.apple.com, but who wants more apps like Apple Music? These are abysmally slow apps.
Things are better on Linux before, but "native" is DE-specific anyways. Between electron, GTK, and KDE, I tend to prefer electron. Only second to apps like Blender or Godot, which have great UX advantages from having their own bespoke GUI frameworks which are consistent across OSes.
It's more that no-one cares about consistency across platforms because each of us only has one platform.
I will never see the Windows version of your app, because I don't run any Windows machines. I will never know if your Windows UI is consistent with your MacOS UI.
I do care if your app uses a bunch of key shortcuts or UI conventions that no other MacOS app uses. If I have to learn a whole new set of UI conventions just because your app decided not obey the standard MacOS UI conventions, that will annoy me a lot.
If you're writing five apps, you're doing it wrong. A huge amount of probably any app can be contained in a common blob of business logic and wiring, with some native presentation on top. Yes, I oversimplify, but folks have been doing exactly that for decades. More work? Yup, even with a common engine it's more work than firing up Electron. That's where "do we even give a shit about what we're shipping?" comes in. If there's no pride in the finished product, well, Electron isn't going to save your product, either.
Especially since all the critical "business logic" in this case that will be shared is likely to be the backend store front APIs and what not. It's the platform things which will end up being radically different.
> A huge amount of probably any app can be contained in a common blob of business logic and wiring, with some native presentation on top.
I keep seeing people claim this, yet most UI software I see really doesn't do anything that complicated. It's some basic database CRUD stuff and some basic algorithms, and then all the actual effort lies in making the UI feel good.
> Yeah the problem is having to write 5 apps, not the inconsistency.
Isn’t that what abstraction and interface is about? You can write the common business logic and presentation concerns, and then share it across each native implementation.
That is what web wrapping is in truth, but it’s just inneficient because of the huge surface area of web technologies.
Aesthetics, usability, and technical flavor are different axes and feel right is a subjective point. And based on anecdata, usability is what actual users want. Web wrapping is mostly an exclusive focus on aesthetics. With native frameworks, you often get very close to "feel right" without much effort.
> With native frameworks, you often get very close to "feel right" without much effort.
I've only seen this sentiment on nerd forums but never among my less technical friends, so I'm a bit reluctant to accept this wisdom honestly. That said, if you're just building something for a technical crowd you'll probably make them much happier if you make your implementation use native widgets and take up much less memory and redraw faster at the expense of a slightly jankier UX. Though, this is probably why IRC clients all feel janky lol.
For example a lot of my less technical friends/family really like the visual design of Spotify even though it conflicts with a lot of design language on MacOS, Windows, and Android.
The hidden reason why so many people are frustrated with 'modern design' that goes unaddressed is that most software nowadays completely ignores the standards of the system it runs on. Every single developer of even the most trivial apps insists that their design language is the most important in the room, which is what brings us to the temptation of using a webapp for everything so every user on every OS sees the exact same thing. It's gotten so bad that the only places I see native system UIs from third parties is either older software or hobbyist projects, but never anything corporate.
> most software nowadays completely ignores the standards of the system it runs on
Good, sticking to design conventions of the OS may not be a good fit for the specific software (e.g. think about a video editor). And even in more-tame cases, people have different aesthetic preferences (e.g. a minimalist text editor).
My software, a video browser / organizer, looks better than anything I've come across (for my preferences). For example, sticking to OS conventions would have required more padding around everything, decreasing functionality - https://videohubapp.com
Some of that is also that what "native UI" means has shifted a bit. To look at Windows, Win95 had a thoroughly considered UI design that was very well adapted to both what it needed to do and the hard constraints on it. Win11 is... kind of a shitshow. (Are parts of it still from Windows 3.1? I haven't tried to track them down... but in Win10 there were a few dark corners.) It made sense to match the native UI in the Win95 days. It often doesn't now, it's simply not a target worth hitting. You could just use Win32 and target the older Win95 look, but apart from appearance issues, it was designed to work well in 1995, and we simply demand different things from our UIs now. Like support for touchscreens. Old Win32 apps are painful on touchscreens. (But, often, awesome without a mouse.) And those are real considerations, not oooh-shiny-new-thing designer BS.
Yes, modern Windows design is a bit of a mess, caused by both the designer bug of redesigning everything (we have had at least 3 different Windows design languages prevalent over the last 10 years) and Windows having to support extremely old legacy software. The second was almost nonexistent when Windows 95 was designed, they could have a true clean slate, so the comparison isn't as fair. But still, even with the abundance of native UI frameworks, the point holds true that no new software uses any of them, no matter if it's Win32 or WinUI. They could implement them very easily, and some hobbyist software does so, and it looks very native and in-line with the rest of Windows 11, but they choose not to.
And yes, the 3.11 ODBC file selector is still in Windows 11. Praise be to legacy components.
As a user of both windows and mac operating systems, I'd rather the app be consistent between them so I don't have to adopt a different workflow depending on which OS I'm using the app on.
Native doesn't mean "not a web view", it means native. The eight Qt apps on my Mac are really obvious, with basic fundamentals like toolbars, drop down boxes, and context menus bearing no resemblance whatsoever to any native Mac app from any decade.
10-15 years ago it was possible to reliably clock a Qt on Mac app before it finished launching, simply from the app using the wrong hourglass cursor. (The old cursor that had once been a hallmark of classic Mac OS apps ported through Carbon frameworks, and which disappeared from daily usage when the Finder was finally rewritten atop the NeXTSTEP-derived Cocoa frameworks.)
Skia is the 2D graphics library used in Chromium. A number of GUI frameworks (Flutter and Avalonia) directly use Skia to do non-native platform GUIs that are consistent across platforms.
Flutter is migrating to Impeller, which has much better runtime characteristics and font rendering. It's already being used by default on desktop and mobile. [1][2] But your point still holds, of course.
Tried using it yesterday via OpenRouter in OpenCode to follow a plan written by a larger model. I didnt find it to be particularly great. It got stuck on a lot of tool calls. Given the price and speed, it would be nice if it could do what it advertises.
Perhaps there are some system prompt tweaks that could get it in line
Been using Flash 4.1 via the ante harness to blast through a GBA recomp. The ante team has pushed hard to make Flash 4.1 perform well under it. So far, I've maybe spent $10 over the last 3 days. Its a real workhorse and works much better in this harness
You are quoting the discount pricing. It is 50% off for the next two weeks. The blog post has the real pricing up front in the card on the right side: https://mistral.ai/news/mistral-large-4/
After that, it will be much closer to GLM 5.3, but you can also get 5.3 in their API! I dont see people really talking about that.
Been a happy Juggler since I first saw it on here. I really think it is much more approachable for people who are intimidated/overwhelmed by the terminal and just want an all in one solution that is intuitive, pre-configured, and also works great. I really like the "auto approve" strategy given lots of people have no idea when something like a `sed` is going to write or read.
However, I likely wouldnt suggest Pi (or even opencode) to someone who I would also suggest Juggler. They just dont feel like the same type of user, in my opinion.
I do think you could take a look at ChatWise, Kelivo, Rikkahub, Jan, and Waku if you wanted to see who your real competitors are. Again, in my opinion.
Those are full standalone GUIs with workspaces and built in tools that help you get things done without requiring any terminal commands.
The biggest win for me is they can have state. So you can log in to an MCP (via oauth) and not worry about having to refresh tokens locally or in the ENV. I use it for Trello. I log in, then I can call all the data about my board. No CLI to install, no token to copy-paste somewhere. It just opens a browser tab to log in when required, next, next, next, and it works.
Your cli is slightly different than how a LLM uses it. We use 'export' and 'cd' to store state. LLMs dont do that generally.
There is an argument for and against having the model repeat this state.
I'm still on team CLI in that i think even designing an interface from the CLI perspective gets you a better domain interface compared to when you can 'cheat' with the MCP state.
But the thing MCP is just better at is credentials.
The thing that _was_ the dealbreaker between CLI and MCPs is that MCP's couldn't be composed. Maybe `codemode` fixed this; haven't tried enough to say 1 way or another.
I also bounce around to lots of different models. Some are cheap and dumb, some not. When unfamiliar with a model, the last thing I want it to allow file edits and destructive tool calls. Having a mode explicitly for that is helpful.
reply