I am baffled by this. I wrote a cross-platform GUI for an audio signal processor in wxWidgets that rendered all of its own widgets (own Draw calls, for faders, graphic EQ, compressors, drag/drop linking of nodes for a layout, knobs, meters/scopes) and incorporated Horde3D as a 3D rendering of the arena. It ran on macOS and Windows. Another chap handled the network control code that included joining multicast groups to discover the devices, and the device itself had a series of DSPs on it with an ARM processor for drawing on the LCD and handling the network stack and interfacing between the DSP and inputs from the network. That's 2 people.
So you're saying that it is impossible for a large company to somehow use native toolkits to draw text bubbles and emojis?? Video and audio is another matter, but MSN Messenger managed 75% of this decades ago, natively.
> So you're saying that it is impossible for a large company to somehow use native toolkits to draw text bubbles and emojis??
One weird thing about software development is that there are plenty of things which motivated individual developers can achieve which large companies can't even write the requirements for, let alone achieve.
> It's not a skill issue, it's an organizational issue.
We can say that for the majority of companies with large teams that already have this. Everyone knows Meta is no different.
In this case, it's a skill issue to ship low quality software which is what whoever that team at Meta just did and knowingly approved.
> The engineers at Meta are world class but they're nerfed by organizational constraints.
That doesn't mean anything given that shipping regressions to billions of users is not of "world class" calibre.
If fact, that is of amateurs behaviour and way below the expectation of a multi-trillion dollar company hiring the "best" engineers which they can certainly afford.
It is a management failure 100%. It doesn’t matter how good you engineers are when they are punished for doing good work and rewarded for shipping garbage.
Hacker News just recreated startups from first principles! /s
But seriously folks, this is why there will always be room for startups.
Megacorps are slow lumbering beats that suffer from entropy and signal loss at every edge on the graph.
The example here is that some marketing function gets the signal from Zuck to put Lego duplo characters everywhere because metaverse. Of course 90% of the company knows its dead on arrival so they are going to drag feet and try to work on their actual problems (or their local feifdom mandate) and wait for the latest fad to blow over.
Super wasteful, super inefficient but hey ads make an insane amount of money so Zuck can do whatever for a loooong time.
This is why Warren Buffet advocated dividends. Companies and people have few good ideas. Management however uses this money from the good ideas to fund boondoghles (see metaverse). The thinking here is that the money would be better utilized returned to investors to fund other good ideas.
> Management however uses this money from the good ideas to fund boondoghles (see metaverse).
I'm not sure we should be critical of risk taking in large companies because there was some notable failures. But maybe it is the sort of thing that doesn't scale well in a large org, they want big immediately.
Warren Buffett would rather have dividends than autonomous driving. Investors like him are the enemy of long term moonshot R&D like deepmind and Waymo. I’m glad Google shareholders don’t think like him
> Investors like him are the enemy on long term moonshot R&D
That's because he comes from a generation who has "the government and universities take care of R&D" as a mindset. A lot of the stuff we take for granted today - *nix, TCP, IP, the Internet itself, lasers, microwave, radar - came out of universities, government grants or the military.
The idea that the government is incapable of R&D is relatively new and originates in small-state / lean-state / starve-the-beast ideology.
And personally, I rather believe the wisdom of an old man who made a metric shitload of money by being a good honest citizen than others who got rich on advertising and stonk market shenanigans.
Imagine having to tell voters that the government is going to spend 11b (Waymo funding since 2020) on automating an entire profession away, good luck winning your next election. It is inherently infeasible to have popular support for the most revolutionary technologies that will cause societal upheaval. Voters have voted to preserve their lifestyles in amber over positive revolutionary change every time, see: public transit, building dense housing, corn subsidies, etc.
> Imagine having to tell voters that the government is going to spend 11b (Waymo funding since 2020) on automating an entire profession away, good luck winning your next election.
So what, NASA spent billions on the new space race, with everyone knowing the true aim was to kick Boeing's ass and Boeing being widely known as a pork distributor first and foremost and rocket/plane manufacturer second.
The job of politicians isn't to cater to the whims of reactionaries, the job of politicians is to do the right thing and sell it to the voter base.
You're proving exactly why this is the case. With 2 people, it's easy to coordinate, agree upon a lot of things, accept compromise, meet face to face. Try 20 people. Infinite meetings, pushback from managers, resource allocation clashes, politics top to bottom.
I'm not defending any large company, they could if they wanted to, they just don't care. If this is "cheaper" and they can cut costs, this is what they'll do.
If they want fungible engineers, yes. They don’t want to pay extra for an expert in wxWidgets and Horde3D. They don’t want to wait while that expert’s replacement comes up to speed on Horde3D. They don’t even want to wait for any existing engineers to learn these two technologies.
They know web, they can manage web - web is “less risky.”
A lot of native toolkits are much more amenable to learning than the web. Web developers tends to create abstraction towers just to say they’re expert in a very small niche. I’ll take autotools and gtk over next.js everytime.
> They don’t want to pay extra for an expert in wxWidgets and Horde3D.
In my experience, in a large software company [0] it's very common for new folks on a "team" to have to spend one or (often) more months learning both the software they're now working on and the various libraries used in its construction.
It's very nice if you can find an expert in everything you've used on a software project, but that's not going to happen often. [1] And -IMO- if whoever you've hired can't generally come up to speed on WxWidgets in a couple of weeks while also becoming familiar with the rest of the project they're now responsible for, they're not someone you want working on your project.
[0] Which is the sort of company we're talking about right now.
[1] In part because companies generally don't want to pay enough to hire such people.
I think what he is saying that native platform apps get delegated to different teams and coordinating among those teams becomes an additional cost. You don't want each team going off and doing their own thing.
Your 'answer' is "use a cross-platform GUI toolkit" but that has its own challenges. Not least that you typically build a native app because it delivers a native experience that users expect.
In general (and I accept there may be counter-examples) cross-platform tools fail to do this.
Is that worse than the usual mobile/web frontend/web backend split? The thing is that there should be more whiteboarding done to design the features. Instead you mostly have product designer trying to impose whatever regardless of the technical challenge/practicality.
Yes. Arguably because of the organisational cost of maintaining both a web version (which they arguably need) and a native windows version (which they arguably don’t need).
Most of WhatsApp is UI veneer on a set of webservices. Unless they’ve architected their app badly, I’m not seeing the cost of maintaining a native software being substantial.
That's not really native either. Whether it's a web wrapper or Electron or wxWidgets, just because something runs "natively" doesn't mean it feels native.
yes, precisely that. Or if you want to generalize further: Companies have staff and processes to "improve" the product, process and productivity. After a few years, when they've done a marvelous job (worth every penny) the job doesn't exist anymore but they continue to tinker and ruin their creation.
In my manual labor occupation the office folk often try to get away with a horribly inefficient schedule designed to make their job easy. I often have to remind them that they work for me. I do the actual work, your job is to optimize it.
This to me is the same thing as replacing a rather decent desktop client with a web wrapper. This is their update, its obviously bad for the customer. The excuse is to make further updates easier. There is reason to think these will also be bad for the customer.
Someone once explained to me that the process of everything we build and create is very similar. If some young sector has a wildly different approach one should be very skeptical. It happens that an architect still has work to do after delivering the drawings. Sometimes it is necessary to recall thousands of cars.
Ideal would be to have a perfect construction drawing then build the machine. The architect moves on and designs the next garden. A plumber puts the pipes in, tests if everything works then goes to the next construction site. The difference with software is that, when done, it can last forever. The opposite of what the industry pretends to be true.
Windows 3.11 also wasn’t shipped to you over a cellular connection when you clicked on it. If it were, 6x1.44MB would have been considered quite unacceptable.
Buy your own domain and run mox (https://github.com/mjl-/mox) on it. In the setup it provides details on the DKIM info you need to put into your DNS records. Get a PTR record from your ISP (if hosting at home); periodically check your blocking from spamhaus etc.
I run mine on a Pi4 no problem whatsoever, but I guess a VPS could also be used, although the scamalytics analysis will show it's a server or an IP shared with an anonymising VPN etc. if it's a shared IP on the host.
Yes. I use it with wxWidgets and other C++ projects, never touching Qt at all. The performance analysis tools on Linux have been useful to me, and the text editor is lovely to use instead of fuzzy-font-land like Visual Studio Code.
What sort of things are most people doing on their desktop computer that needs more power or RAM though? I can't imagine.
You can still buy woefully underpowered laptops with hopeless resolutions and with 4GB of RAM running Windows 11, and that is a horrible desktop experience. At least with this it is a usable desktop machine, where the normal bottleneck was IO speed.
This is an excellent summary of the issues I have faced, where if you inform someone in the middle of a conversation of an experience you had in the past where someone else has done something bad, they tell me I am being negative... But it wasn't me who did the bad deed. It is completely baffling to me, and you summarised very well how people just want to effectively be tickled instead of dealing with facts. It is very isolating.
I noticed on lesser devices that their word-wrapping algorithm is very very slow, so inputting a large block of text in a native Android text box got slower and slower and slower and slower.
I have been using code-server (openvscode-server will also do) to run on a remote powerful machine and then connect to it from lesser devices (a phone in desktop mode would be good if I could do it!) via a VPN. Keeps all the source code remote.
So you're saying that it is impossible for a large company to somehow use native toolkits to draw text bubbles and emojis?? Video and audio is another matter, but MSN Messenger managed 75% of this decades ago, natively.
reply