Did we lose our way in making efficient software?(rufatmammadli.medium.com) |
Did we lose our way in making efficient software?(rufatmammadli.medium.com) |
While I've been involved in releasing both types of software (a ye olde Windows standalone app written in Qt framework and the new software that's being released every 2-3 days or so) I find the new way much less painful. I couldn't imagine releasing a Qt app in such a cycle (how to do updates? maybe like Minecraft - every other launch there's a new package of 64MB downloaded)...
...but as a user, I feel much more comfortable when code is on my computer, available for me to run. While extensively using Google Docs to share documents with my wife (calendar, spreadsheets) and band (songlist, lyrics), I'm scared that if I moved everything to on-cloud (paid mail, photo enhancing (thankfully I'm a standalone Lightroom buyer), vector graphics (inkscape user), photoshop (gimp user), etc.) one day I would hit the roof in payments of my per-month-but-bound-to-year plans. In 2030 it would be also netflix, shmetflix, televisionix, phonix, car-as-a-service, heated-seats-as-a-service, air-conditioning-as-a-service, toilet-as-a-service, smartwatch-as-a-service etc. building up to an unbearable rent for something that used to be free or paid once. I don't feel like my duty is to provide a constant supply of money to my beloved corporations
It looks like a joke seeing an i5 choking on files opened by PC's from 2003.
No.
Right?
- how much computing power is needed to present a single useful bit of information to the user;
- how much computing power is needed to process a single useful bit of information;
- how much total data transfer is needed to transfer a single useful bit of information.
of course to answer the above questions you need to give the definition to the term "useful single bit". And the hint here is: if we agree that - say - rainbow has 7 colors then the information about all 7 colors would take just 3 bits of data, wouldn't it?..
long story below:
this questions pops up at least twice every year by someone completely frustrated by the current state of things in the computer industry. And if you think it is limited to only software side of things, then well... ignorance is bliss.
Think of it from the incentives and rationale perspective.
Whenever you encounter a bloated piece of software or an over-engineered hardware box put yourself into the shoes of the author of this. Once you delve into the details of why and how was any specific tool or technology created you can understand why it looks so bad.
Some notorious examples of most hated programming languages were created in a very short period of time without any strategic thinking involved by people who had not had any experience with designing programming languages.
And people continue with this pattern in all types of software driven by business requirements rather than their engineering and scientific aspirations and talent (or lack thereof) most of the time.
In other words, bloated software is the result of time limitations imposed on developers. Efficiency, size, quality, stability and security go out of the window when you need to pursue other, more "important" goals.
"We need to go ahead of the competition and the time to market is our priority. We'll cut corners and burn cash. No thorough think through, just do it."
Another perspective is resource limitation. You as a developer have access to virtually unlimited computing, networking and storage resources. Remember this "memory is cheaper than developer's time" mantra?
Now put yourself in the shoes of a NES game developer. You need to squeeze in the whole universe into 32kb, with graphics, music and gameplay that will look attractive and responsive running on a 1.8MHz single-core 8-bit CPU.
Or put yourself into the shoes of the Voyager 1/2 team whose objectives are to keep a small piece of metal afloat in hostile environment for the next 50 years. With remote debugging capabilities, over-the-air software updates and continuous telemetry transmission back to Earth.
If Brendan Eich had not been given just 10 days to draft the javascript specs would we see something different in the frontend world today? Or we'd still see the 10MB garbage being downloaded by every other website just for the sake of keeping the cables busy?
And here is one of my favorite quotes by Alan Kay:
"Think about it. HTML and the Internet has gone back to the dark ages because it presupposes that there should be a browser that should understand its formats. This has to be one of the worst ideas since ms-dos, this is really a shame. it's maybe what happens when physicists decide to play with computers."
hiring bar was dropped. expecting a mid level engineer to work with a byte buffer is considered “too complex” and non differentiated work.
the literal goal is to pump out features written up by mba/product team. none of these mbas use the product mind you. theyre chasing stupid features they think vice presidents want, because the thinking is it will drive promotions.
this is a cynical post and i will stop here. my org has problem of incentives. nothing else. you incentivize wrong things then this happens
(probably the cost of making an ancient Word and the newest one... : )) but there could be lots of other examples of 'modern' ones made along current trends vs. similar feature set classic ones, I wonder how this cost characteristic would play out .... : ) )
Moore's law kept going, and software started getting a little bit faster, which was enough to stop undoing the gains made by hardware, and now things are back to mostly snappy.
Occasionally you'll get a 30mb file that's slow... but subjectively things sure seem better than 10 years ago when you couldn't even think about optimization without someone beating you over the head with a "premature optimization is the root of all evil" quote.
This is the inverse of my experience. There are few applications that have a UI that I would refer to as 'snappy'. In fact, I am trying to come up with a single example, and atm I can't even think of one.
Most mobile sites in general are fast, maybe not "2kb of HTML" fast, but fast enough I don't notice or think about performance when browsing.
Although, I suspect a lot of programmers can think faster than average and seem to be bothered more by small delays than the rest of us, since they're able to do things like figure out a Vim keystroke sequence instantly.
If it doesn’t matter - it doesn’t matter. If the goal is making a document format that is flexible enough to accommodate history, concurrent editing, various layouts and embedding, etc, all this comes with abstractions that add inefficiency. The trade off is ease in adding and changing to the format and the software that consumes and produces the format. If the consequence in the real world is effective unobservable in any material way, who cares?
Maybe as a moralistic measure it’s offensive that something lacking parsimony is practical. But from any meaningful measure - the users perspective, the developer, even the company paying for the processing - if it doesn’t matter - it literally doesn’t matter.
Comparing Google Docs to a program hosted on an Apollo era flight computer is obtuse to an extreme, and I would rather write my collaboratively edited documents with Google Docs than Apollo era flight computer any day no matter whether one is less parsimonious than the other.
Except the post you're responding to was literally in response to a user problem trying to edit a 30MB document in Google Docs. So it very much does matter, from the user perspective.
> Comparing Google Docs to a program hosted on an Apollo era flight computer is obtuse to an extreme, and I would rather write my collaboratively edited documents with Google Docs than Apollo era flight computer any day no matter whether one is less parsimonious than the other.
Straw man. The post compares Google Docs to LibreOffice (a competing product), and points out that LibreOffice solves the user's problem (editing a 30MB document) and Google Docs cannot.
Again, off the top of my head I can't think of a single application I would describe as "snappy". Or, put another way, I can't think of an application that, in recent memory, really impressed me by how well it performed.
https://github.com/jmechner/Prince-of-Persia-Apple-II https://www.jordanmechner.com/en/books/journals/
When I went to recreate it, I contacted QNX and asked if I could speak to the guy who did the work, and he had died the year before. So, I just took apart his floppy, and figured out how he did it.
The things you can do when you invest your time 100% in something.
Today’s software systems are more generalized, though they are solving the same business problems, just with more details / functionality than before.
Sad for me who wanted to be a software architect. I had to watch all this unfold in real time from inside various companies and never had the ability to fix the problems. Last time I tried to prevent major architectural flaws from being implemented in software during the design phase, I couldn't convince management and had to quit the company... Then 2 years later, from the outside, I witnessed the project turn into a complete failure. They literally abandoned the whole thing and started using a competitor's platform... Which, to rub salt into the wound, is almost just as awful.
Typically the "architect" title I see thrown around today pertains to gluing together some abominations of "systems", typically cloud services to do web stuff at "scale"
Architect today seems == "cloud infrastructure decision maker" and has no bearing on the code written, libs/frameworks/whatever used
Why didn’t you just, like, get him Word? Why did you make him try to use a shitty web app that assumes everyone’s computer is brand new, then install an open-source program that’s going to be constantly playing catch-up with Word’s updates and may cause problems down the line when Dad wants to work with someone else’s Word docs?
Maybe there was a perfectly good reason for this choice. I can think of a few. Maybe you helped Dad enter The Year Of Linux On The Desktop recently. Maybe Dad didn’t want to pay for Office. Who knows. Whatever the reason, you didn’t put it in this post. And you ended it with a plug for your completely unrelated SAAS, too.
Update: I updated the article.
Now that the zero-interest economy is over, the entire tech sector is readjusting.
I'm not going to deny that we could do better, but it is more nuanced than that:
OP uses Word as an anecdotal example, but Word is not designed with a goal of being optimized. It is designed with a goal of being backwards-compatible to decades of history.
We cannot assume that all software shares the same goals because they simply do not. When we look at the problem any given software is trying to solve, performance optimization is almost always important, but almost never #1... #1 is "solve the problem". Doing it fast is always secondary to doing it at all.
Also, Word (despite the general fame) isn't compatible with decades of history (remember a post that found that LibreOffice was more compatible with history (though not current designs))
Once all that XML is naively loaded as nodes, we might be talking about more than 1000 MB of RAM usage.
Did you consider trying Microsoft's own browser-based Word editor? It's free too. And .docx is it's native format.
Or, consider doing a conversion to Google Docs native format first (you'll lose some formatting though, possibly a lot of it).
[0]: https://ecma-international.org/publications-and-standards/st... [1]: https://en.wikipedia.org/wiki/Office_Open_XML#Application_su... [2]: https://www.theverge.com/2019/4/10/18304978/google-docs-shee...
It is just a different kind of efficient, there is economic incentive in making software development process more efficient and not that much incentive for making software itself efficient .
Software development is more accessible to millions and millions of new developers due to the work on higher level languages, frameworks , IDEs , libraries , low code , copilots and so on .
Each of these innovations made software development more efficient (not necessarily faster ) .
Nobody buys or uses software because it is faster , only cheaper .
Now it is merely not elegant or not as fast as it could be or at least fast enough initially during development and so no effort is spent on making it better.
I believe this could only be fixed through regulations which either make the engineers liable (thereby empowering them to make decisions) or by regulating energy use and user experience.
If we don‘t want this, then things will just be the way they are now.
But what are companies doing with this fast turnaround time? Features suck and are largely incomplete in modern software. For example: Sonos speakers, if the WiFi goes down they don't reconnect. Why? Why is basically every device and every bit of software chock full of obvious stuff like that? Do we really need AI or something to tell us how to build something properly?
Anyone that can't be bothered to update their ten year old laptop because it's slow is also not going to spend a lot on faster software. There's just not a lot of money in optimizing stuff. And if you have a modern laptop, it doesn't really benefit a lot from the type of optimizations you'd do to make things run smooth on a ten year old laptop. Especially when optimizations are simply about turning stuff off that aren't really in the way on the faster laptop. Like having some cheap 3D effects, pretty colors and animations, etc.
Anyway, I'm old enough to know that this is not a new debate. We never lost our way on this front. It was always like this even when computers were several orders of magnitude slower.
Also, cloud-based synchronization using CRDTs is a complex problem that is significantly more complex than just loading the document.
Can't claim we are going backwards when comparing apples and oranges.
It uses 500 Mb of ram fresh, and in a couple weeks it goes up to 1.5 to 2 Gb and I have to kill the tab and reopen it.
This is the modern javascript world...
There are some new simplified approaches that are starting to be interesting again.
Also, in the past you had to care explicitly about how much memory you allocate etc. which stopped you to think. Now you can pretend you have infinite resources because everything happens implicitly.
Compounded with this [0]:
> O(n^2) is the sweet spot of badly scaling algorithms: fast enough to make it into production, but slow enough to make things fall down once it gets there
you get what you get. Ever opened a GitHub pull request with 2000+ files changed? It hangs the M1 MBP. The solution is probably not rocket science, if someone really prioritized the fix.
[0] https://twitter.com/BruceDawson0xB/status/112038140670042931...
1. JS doesn't support multithreading, nor many other features that are useful for performance (e.g. mmap). This severely limits what you can do and makes it hard to scale up by parallelizing.
2. JS is a very pointer heavy language that was never designed for performance, so the CPU finds it harder to execute than old-school C++ of the type you'd find in Word. It's hard to design tight data structures of the kind you'd find at the core of Word.
3. The browser's one-size-fits-all security model sacrifices a lot of performance for what is essentially a mix of moral, legal and philosophical reasons. The sandbox is high overhead, but Docs is made by the same company as Chrome so they know it isn't malicious. They could just run the whole thing outside of the sandbox and win some perf back. But they never will, because giving themselves that kind of leg up would be an antitrust violation, and they don't want to get into the game of paying big review teams to hand out special Officially Legit™ cards in the same way that desktop vendors are willing to do.
4. The DOM is a highly generic, page oriented structure, that isn't well suited for app-like UIs. As a concrete example Chrome's rendering pipeline contains an absolute ton of complexity to try and handle very long static pages, like tiled rendering, but if the page takes over rendering itself like Docs does then all this just gets in the way and slows things down. But you can't opt out (see point 3).
Exclusively the minds of developers and the stance of management.
It is of course possible to built responsive, high quality and performance websites. But that is hard, much, much harder than to make something work, which maybe takes a few seconds to load, sometimes doesn't work quite right and can be a bit tedious to use.
Also interoperability I guess is a problem (html really IS a consistent UI for every platform)
Also, Docs uses a custom UI framework already. It implements all the UI controls itself, the browser is only really used for rendering text and styled boxes. I remember the first version of Docs when it was called Writely used the browser's built in editing support but they had to abandon it because it was too buggy, so they moved to using JS to lay out every character and draw their own cursor. It was considered wild and crazy at the time but Chrome was getting fast enough to make it work. Of course, it's more efficient to have editing be implemented fully in C++ but browser makers never managed to make that work properly, so, slow path it is.
The Apple Developer Program is only needed for macOS if you want to do sign your binaries or distribute through the Mac App Store. And you only have to pay Microsoft if you want to publish to the Microsoft Store (or use Visual Studio if you're a company that has more than 5 Visual Studio users, more than 250 computers, or more than $1 Million USD in annual revenue).
> buy certificates for signing binaries
Fair (though both Windows and macOS will run apps that haven't been signed, with more warnings of course).
> share 30% of my revenue with them for barely any reason
Only if you use their stores (Mac App Store or Microsoft Store), and it looks like the Microsoft Store won't take any cut if you do your own payments and it's not a game.
As for, Apple, I do not know but I suspect you can make Mac applications without a developer account. You need a developer account for iPhone. It's $99 a year the last time I looked. This is not a lot of money if you are serious about making an application.
Compare with the web where LetsEncrypt just works without demanding a king's ransom.
As for the APIs, it is very easy to get into dependency hell between all the different UI technologies, .NET implementations, and target systems. Want to develop a brand new plain-old GUI app? Probably simple (although I've never tried, the web is right there). Need to develop a plugin for an existing application, or a new app for something like Hololens? Have fun.
It is a lot of money when you consider it should be free and serves exactly no purpose.
Maybe not for you.
Set up CC processing on the web:
How much are you going to pay stripe? 2.9% + 30¢ ... that means you have to charge 10 bucks to get down to a 6% transaction fee. Quite the price floor and an interesting cap on your pricing model!
What does managing chargebacks cost you? The moment your taking money your going to hire in customer service, or spend time dealing with CS. What happens when you get a chargeback, or do a refund? Most of the time you loose money (processing fees etc)
If your under a million bucks a year apple is 15%. If you're building a low price app or a value add app, odds are that apple is going to be a far better deal for you than doing it on your own.
Chargebacks = customer support. I agree with that, but if you have a B2C business which has any non-trivial revenue (OP is talking about word doc apps, so we’re obviously not talking about indie $2 side project apps), then you would already have CS anyway. I fully understand there is an opportunity cost with any service and where those costs get realized, but your examples don’t seem like a slam dunk in apple’s favor.
How is chargeback being managed on Apple? I doubt they are swallowing the cost on their side, so I don't really see the difference between what'd get with a bank: you're losing the money anyway.
And once your rate goes to 30%, does it stay there the following year, or does the whole system reset to zero each year?
How do you calculate a price for not being able to release your main product? Usually without clear indications of what exact interpretation of a rule you are breaking...
We've had delays of a week because of things like we mentioned "Android" in an integration setting that had been there for years.
?
Your math seems to show the exact opposite.
I'd happily build iOS apps without XCode or any of Apple's frameworks to save the 30% fee. Heck, I'd do it even if I still had to pay the 30%, I hate being forced to use XCode.
That said, I don't know about Mac, but you can build apps using free tools - maybe in not as convenient way, but certainly you can.
I remember, because I was someone who couldn't afford Visual Studio licence and had to make do with GNU tools.
The greed of these companies put me off from developing anything.
Does the App Store collect sales tax and remit on your behalf? If it does then I think it's worth it or face registering both in the EU and UK ($0 tax threshold) as well as 50 US states (once you hit the allowed limit) will take you a long time.
Apps like that get made anyway but as it stands at least there’s a healthy crop of smaller/indie native alternatives which often best the behemoths in UI/UX. That would likely disappear with the addition of a standardized UI API, as it would probably also come with the abandonment of the old specialized APIs.
[1] Yes, one can use Qt for commercial software without buying a license (as long as it is dynamically linked), but their marketing does everything it can to hide that fact. Also, the newer additions to Qt do not fall in this category – for those, you have to pay.
That's not realistic for Apple users who are used to ergonomic software. It's not technically required to notarize, but practically speaking, it is.
Choosing overly complex web frameworks is still a guilty pleasure of too many projects.
This is an important subject, thus it's one for which clickbait is generated.
Size is a problem. I look at my Rust compiles scroll by, and wonder "why is that in there?". I managed to get tokio out, which took some effort. The whole "zbus" system was pulled in because the program asks if the user is in "dark mode". That brought in the "event-listener" system.
Lately, "bash" in a Linux console has become much slower about echoing characters. Did someone stick in spell check, or a LLM for autocomplete, or something?
Obviously your statement is true about most other sites, but I thought it was an odd thing to say about a platform that famously doesn’t serve ads.
Software has been freeriding on hardware improvements for a few decades, especially on web and desktop apps.
Moore's law has been a blessing and a curse.
The software you use today was written by people who learned their craft while this free-ride was still fully ongoing.
Developers certainly like to have their completely integrated, connected and universal computing platform (the web). And users do not seem to particularly care about performance as long as it is good enough. And that is exactly the standard that is set, software is allowed to be so bad that it doesn't really annoy the user too much. Management doesn't care either, certainly creating good software isn't important when good enough software has already been developed.
Sure, I would like things to be different, but until one group decides that a drastic departure is necessary, nothing will change. There are also no real incentives for change, from any perspective.
The reality is that these are all business decisions:
1) Move to the cloud because the business likes the steady payout of subscriptions. Business customers love not having to hire IT teams and demand six 9s of uptime because it is someone else’s responsibility. But performance needs to just be acceptable to end users.
2) Customers refusing to upgrade on-premises software, that led to long maintenance cycles and endless patches
3) Developing once for the web vs. Multiple times for different platforms – each needing its own developers and testers.
No amount of expertise on the part of developers is going to address these fundamental forces.
After a certain period of time, that software worked just fine for those customers. Photoshop is a great example. Sure, you won’t get the flashiest features, but CS4 will still work for you on a Win7 machine without any additional fees paid.
It did pretty much everything it does now, only lacked a grammar checker. (WordPerfect had one.)
Now we measure things in GB units. 1000X bigger, but what was gained?
We not only lost the way, we don’t even know the destination any more.
In terms of frontend, which the post focuses on (Google Docs and a 30MB doc), I guess I'm conflicted. While I tend to favor native apps + web pages, I'm also a daily Tiddlywiki user, and I really think web apps have their place (heck, one idea I'm working on is a lightweight local server that lets you run web apps like Tiddlywiki). But without a doubt, Tiddlywiki is more resource intensive than Emacs (my go-to for notetaking when I'm not on TW). My tab for a 6MB Tiddlywiki file uses 155MB of RAM, and my (heavily customized, dozens of open buffers) Emacs session uses 88MB. So I do think the author has a good point.
[0]: https://fennel-lang.org/ [1]: https://sqlite.org/althttpd/doc/trunk/althttpd.md [2]: https://fossil-scm.org/home/doc/trunk/www/index.wiki [3]: https://makoserver.net/
1. Company executive decides their developers need top-of-the-line hardware to remain competitive in today's market
2. Developers make web apps on their company-provided M5 Ultra Pro Max 128GB RAM powerhouse laptop
3. They never test it on their father's old 2010 family PC, or at least they don't test often/thoroughly enough to realize many parts are broken or unusable
I develop using Dev Tools set to mobile and throttled connection. This way mobile-first responsiveness (limited screen-estate) and potential problems with bad connections are first class citizens.
Now, what usually happens is that I signal problems to the product owner and they wave them away. So I might update your third point that sometimes they do test it on their father's old 2010 family PC but it's not of concern to more relevant stakeholders.
I used to do the same on iOS, but came to find that performance differences on older devices there generally weren’t nearly as severe and that iOS users as a whole tend to use newer devices. When combined with reasonably well written Swift, performance on old devices generally isn’t a problem.
It was suggested to only display the first 100 items and let the user type in 3 characters until it started rendering.
Unfortunately this is the reality for many these days.
Of course instead we just fixed the shitty react code and it rendered instantly.
If the new frameworks make the problem blindingly obvious so that someone can actually justify fixing it, all the more reason to use those frameworks.
As long as they’re sorted and I can jump with the keyboard, that bare-ass drop-down is probably going to “just work” with default behavior. Anything further and we don’t know the intended use case for the element itself, but on the surface… it could be fine.
We used to have developers who took less time and wrote better code.
A surprising number of inexperienced developers do the following: "once I get any working solution I should immediately open a PR" and let the senior engineers tell them what's wrong with it.
When the big money leaves this field I hope there will be more pressure for people to adopt good engineering practices. I love to work with folks who put good effort into trying to make high quality changes. Personal initiative and ethics are how high quality software gets written.
When I compare apps from 2000 to now it's a general decline in responsiveness and resource utilization.
This is flagship product of one of the largest companies, and even they cannot get UI performance right...
The rest are simply modifiers on that value. More intuitive UI allows users to gain value more efficiently. Performance allows users to gain value more efficiency.
Efficiency is important, but Value more so.
The rest of us and the vast majority of professional SE work for a marketing/sales person and “number of new features released this Friday” as you said.
"The classic response to accusations of bloat is that this growth is an efficient response to the additional resources available with improved hardware. That is, programmers are investing in adding features rather than investing in improving performance, disk footprint or other efforts to reduce bloat because the added hardware resources make the effective cost of bloat minimal. The argument runs that this is in direct response to customer demand.
It is most definitely the case that when you see wide-spread consistent behavior across all these different computing ecosystems, it is almost certainly the case that the behavior is in response to direct signals and feedback rather than moral failures of the participants. The question is whether this is an efficient response.
The more likely underlying process is that we are seeing a system that exhibits significant externalities. That is, the cost of bloat is not directly borne by the those introducing it. Individual efforts to reduce bloat have little effect since there is always another bad actor out there to use up the resource and the improvements do not accrue to those making the investments. The final result is sub-optimal but there is no obvious path to improving things."
Web pages/applications are probably even worse in this regard because I'm not sure users even conceptualize them as using resources on their local computers, so they don't get blamed for it (people seem to attribute resource usage only to the web browsers, not the sites themselves)
Some will argue that it is indirectly benefitting the users who can get more features quicker. But most people care more about stability and not having to upgrade their computer yet again than features.
Gonna need some data on that assertion, since there is surely some "balance point", that probably depends on the industry/software, it's not all to one or the other.
As a personal example, these days I've been using Apple Pages (or whatever is called) and it crashes around once in an hour. But it has some features that allows me to quickly iterate over some document, so I am back to cmd+s as in the old days, vs using LibreOffice interface where it would take me an estimate 2x-5x more time.
Try to upload a folder with 20+ small files(say a picture galley) it takes a lot of time to process and upload them. If you add a new file to the folder and try to upload the folder, it will need to upload the whole thing again.
It was slow as molasses - unusable, really, and we're not talking about huge amounts of data either.
At least then, it seemed to be the google drive i/o that was the bottleneck, and the solution was to upload the training files to the colab session / VM.
They just open up "internet" and work on docs, and for 99% of the cases Google Docs works fine despite running in a browser that is much less efficient than a native "app". For most cases it's more than enough for the regular user who is used to "computers being slow" anyway.
By then, the app is built and running. Even though the code is a mess because the developer only know about React, nothing about the DOM and software architecture.
1) Your browser is always open, whereas you need to close your current app and open the app store app
2) Google Search is better at giving you what you want in the fewest number of keystrokes than any search from any other company including app store search boxes
3) installing a "web app" is one click after the google search results or if someone posts a link. As you might know from social media, even one click is a lot and most people won't click links in a comment. In the apple app store i have to tap on the app in the search result, tap install, double click the lock button, scan my face then wait many seconds. Some websites load in many seconds but that's not that common and it's considered a bad website
4) software is reinstalled on each use, meaning it's always updated. Native apps update randomly with a big delay and I have to check if there's an update manually sometimes
5) with native software there's a risk it will not support your device. Risk is commonly expressed as a cost, so the fact that in my life I've searched for an app and found out that it's not supported on iPad or not supported in my country can be counted as every single app store download since then took me .1 seconds longer because I now know there's a risk I will waste my time looking for the app. Web apps also don't work sometimes but it's more predictable since it's usually tied to real limitations like the screen size of your device. Another risk is malware, I feel nervous and powerless when installing native software because I don't know who I'm giving access and to and to what, whereas I understand what web apps can track. and also the fact that it's HTML and JavaScript instead of opaque assembly instructions makes ad blockers possible and cheap
Installation is an integral part of using software, and a big the reason the web won and continues to win.
Then I got uBlock Origin to turn off JavaScript, remote fonts, and large media items.
Result: 116KB
So 98.61% of the page is extraneous...
1. Why bother optimizing when the developer's time is more expensive than RAM and CPU power? I see this a lot.
2. From the times that I can remember (mid '80) till now, only top developers write efficiently software that is efficient. Most developers are average (this is not bad, it is just an observation) and for the average developer software optimization is too expensive in terms of time invested. Some don't know how to do it, some are not proficient enough to do it in the constraints of the projects given to them by bean-counting managers. "good enough" quality is software management is much safer than "good enough" Boeing planes, so when Boeing is cutting corners then managers of developers cut even more.
The comparison should be between developer's time, and time spent (wasted) by all users combined. This depends on # of users, and how often they run the software.
For a one-off, with a few dozen users running it occasionally, yes developer's time is expensive.
For popular software with 100M+ or billions of daily users, developer time is practically irrelevant, and spending weeks/months to shave off 1/10th of a second for each user's run, would be a no-brainer.
Most software sits somewhere in between.
But... developer is paid by company not by end users. And company cares about other things than the interests of society-at-large.
So it's mostly a case of bad incentives. Companies don't care about / aren't rewarded (enough, anyway) for saving end users' time. Open source developers might, but often they are not rewarded, period.
"... the cost of bloat is not directly borne by the those introducing it. Individual efforts to reduce bloat have little effect since there is always another bad actor out there to use up the resource and the improvements do not accrue to those making the investments."
This is the obvious low effort, low complexity solution. Of course you could make it fast, but that would take time and effort for a feature most people won't notice.
The new Passwords is a joke. UX errors all over the place, modal based view with a toggle to start editing. If you need to enter a password in another area of System Preferences, you need to back out of the auth flow, switch to Passwords and copy the credentials over to a temporary file.
So when my doctor sends me a one page checklist of how to prepare for a procedure, I have to open it in a powerful word processor and since it's I'm not using MS Word, the fonts and formatting aren't as expected.
On Windows, Wordpad was plenty enough for most needs, came preinstalled for free and barely consumed any resources, but I understand it's no longer shipping with Windows. Office 365 is now where the money is, even for basic needs.
Find the biggest CAB file, it weights lots of MB. Extract it. Copy Wordpad.exe anywhere else. Delete the rest if you want, or enjoy Sol.exe and friends.
Once the beancounters at the rent-seeking companies (Apple, Microsoft, …) have figured out that web development is where all the money is, this will change rapidly. Google has already started gatekeeping the web via Chrome.
Considering the whole point to have Windows is to use apps I'd expect they made the process super smooth.
I'm old enough to remember when buying development tooling for DOS or Windows was $$$$$$
Today Apple is taking percentage of every dollar made from application developers who participate in their App store and they are making it increasingly difficult to avoid this with every release. IMHO, they are making far more dollars today than they ever did selling development hardware and SDK licenses.
I wish Mozilla or Google or someone aggregated statistics for cpu/memory/energy usage by domain to shame devs who clearly don't otherwise care.
The web feels like 2005 again. Only thing is, this time the popups are embedded in the page...
Now just a decade later, a computer with less than 8Gb of RAM is unusable. A computer with 8Gb of RAM is barely usable. Each new software uses Electron and consumes roughly 1Gb of RAM minimum! Browsers consume a ton of RAM, basically everything consumes an absurd amount of memory.
Not talking about Windows, I don't even know how people can use it. Every time I help my mother with the computer is so slow, and we talk about a recent PC with an i5 and 8Gb of RAM. It takes ages to startup, software takes ages to launch, it takes 1 hour if you need to do updates. How can people use these system and not complain? I would throw my computer out of the window if it takes more than a minute to boot up, even Windows 98 was faster!
Addition: consider also how few resources these applications used, & how they, if they were able to run natively on contemporary systems, would have minuscule system demands compared to their present equivalents with only somewhat less capability.
I think that is some kind of fallacy. We are doing the same things but the quality of those things is vastly different. I collect vintage computers and I think you'd be surprised how limited we were while doing the same things. I wouldn't want to go back.
Although I will say your experience with Windows is different than mine. On all my machines, regardless of specs, start up is fast so the point where I don't even think about it.
I pulled down an Audiobook player the other day, once all dependencies were meet, it need 1.3GB to function! At least VLC is still slim.
> I would throw my computer out of the window if it takes more than a minute to boot up, even Windows 98 was faster!
Sure, Windows has grown a lot in size (as have other OSes). But startup is typically bounded by disk random access, not compute power or memory (granted, I don't use Windows, if 8GB is not enough to boot the OS then things are much worse than I thought). Have you tried putting an SSD in that thing?
(And yes, I realise the irony of saying "just buy more expensive hardware". But SSDs are actually really cheap these days.)
Well said. I believe many of the "hard" issues in software were not "solved" but worked around. IMO containers are a perfect example. Polyglot application distribution was not solved, it was bypassed with container engines. There are tools to work AROUND this issue, I ship build scrips that install compilers and tools on user's machines if they want but that can't be tested well, so containers it is. Redbean and Cosmopolitan libc are the closest I have seen to "solving" this issue
It's also a matter of competition, if I want users to deploy my apps easily and reliably, container it is. Then boom there goes 100mb+ of disk space plus the container engine.
It's really only Linux where you have to ship a complete copy of the OS (sans kernel) to even reliably boot up a web server. A lot of that is due to coordination problems. Linux is UNIX with extra bits, and UNIX wasn't really designed with software distribution in mind, so it's never moved beyond that legacy. A Docker-style container is a natural approach in such an environment.
As long as we have COMPETITION as the main principle for all tech development — between countries or corporations etc. — we will not be able to rein in global crises such as climate change, destruction of ecosystems, or killer AI.
We need “collaboration” and “cooperation” at the highest levels as an organizing principle, instead. Competition causes many huge negative externalities to the rest of the planet.
Cooperation with no competition subtracts all urgency because one must prioritize not rocking the boat and one never knows what negative consequences any decision one makes might prove to have. You need both forces to be present, but cooperation must also be the background/default touchstone with adversarial competition employed as a tool within that framework.
Every modernization (hardware and framework) in software is a tax on the underlying software in its functional entirety
Great take. It feels like the path of least resistance peppered with obscene amounts of resume driven development.
Complexity in all the wrong places.
It wasn't supposed to be like this but it looks like most people never have found the way by now.
So, misguided efforts, wasted resources, and technical debt piles up like never before, and at an even faster rate than efficiency of the software itself declines on the surface.
We use JITs and GPU acceleration and stuff in our mega frameworks, and maybe more importantly, we kind of maxed out the amount of crazy JS powered animations and features people actually want.
Well, except backdrop filter. That still slows everything down insanely whenever it feels like it.
If you feel that people don't care about performance and download size, you may be asking the wrong people the wrong questions.
This is not exactly a new phenomena. People have been complaining about software bloat since at least the mid-1990's. I suspect someone older than myself would gleefully explain that the complaint's went back to the mid-1980's, mid-1970's, etc.. Eventually it gets to the point where only outliers will complain. Everyone else will simply upgrade, put up with the bloat, or stick with old software.
Then there is the question of whether the bloat is worth the benefits. If Docs was a simply a clone of Word, few people would have adopted it. Some people use it because it is free, others since they want to work on or access their documents from various devices, yet others want to collaborate on documents seamlessly. If you're getting something out of the bloat, you're less likely to think of it as bloat.
We also have to consider that some bloat isn't really bloat. It's easy to point to AppleWorks on the Apple II then bemoan how modern word processes require about five orders of magnitude more resources, while ignoring how resource intensive the niceties are. Want to use proportional fonts that look nice at any size and have them rendered properly on the screen? That's about three orders of magnitude more video memory, additional CPU and use for rendering the text, etc.. I'm using that example since it is something people can actually see. Now consider the things they cannot see (such as working on documents larger than the computer's main memory, the memory required for Unicode fonts, the ability to switch between the working document and research notes, memory protection to prevent an ill behaved application from wiping out all of your work). Yes, bloat exists. On the other hand, a lot of the increased resource use is actually quality of life improvements.
Apps today are no more bloated than they were last century, while we gained a lot of functionalities that would have been considered witchcraft in the days of Win98.
Web developers do of course, but I've hardly touched web development myself. Web interfaces etc., are a choice, but I think it's driven by commercial needs-- a desire for subscription revenue instead of one-time sales, etc.
Much of the modern cloud-based or half-online world is quite unnatural from a user perspective, and where there is no need for monetisation-- for example with OpenOffice, the software can expected to remain a desktop application.
This is certainly a big part of it, but it's not the whole story.
For one thing, there were ways to achieve those business models with native software too - "web-based=subscription" isn't actually a requirement, or the only way to go.
But users in the early 2000s, many of them more technical than users today, rejected this idea. It felt like native apps had to cost money one time, and doing otherwise would be wrong. But with Web, since users understood that it was hosted elsewhere, it "made sense" for it to be a subscription, so users went with it.
This also affected the technical things as well. Auto-updating was incredibly frowned upon in the 2000s - you bought it, you got to keep it as-is. So companies had to work very hard to keep multiple versions working all the time.
Most of these biases by technical users have gone away. We now have auto-updating subscription native apps, e.g. Photoshop works that way today. But these technical biases drove the usage of the web, because it was so much technically easier for businesses, and allowed much better business models.
(And, and this isn't even getting into the whole "installing software was really hard for users" thing!)
Nobody complained about that. In fact, few people complained about a few portions of the app that had abysmal performance. It often wasn’t until 60 second load times that customers started complaining.
They still raved and raved about that software because it solved an extremely valuable problem for them. A job that took literally a week could now be done in minutes.
As the space heated up, we needed to improve some things, but most people literally did not care. It would always be stack ranked last in the list.
The problem with the long startup is that it tends to cloud any discussion on performance. Code loading and parsing is basically the biggest bar in the app-perf breakdown of your profiles, and thus spins this narrative that this is the thing to optimize for, because it's the biggest bang. Rather than say, responding to user selections, reducing jitter and sluggishness while scrolling, etc...
I'm starting to believe that for a large class of apps, developers should look at it as if they are writing video games: the user will tolerate the spinner before the level, but then it needs to be silky smooth after. And the _smooth after_ requires a whole class of other optimizations; it's striving for a flat memory profile, it's small ad-hoc data transfer, it's formatting data into usable layout at lower levels in the stack, it's lazy loading of content in the background, etc... Those are the areas where web-devs should be looking a
(again, this only applies for that sort of SPA; e.g read-only content, blogs and such, should display _fast_).
This is a big point isn't it.
We seem to think that customers are choosing "slow" over "fast", when a lot of times they are really choosing between "slow" vs "manual" (i.e. very very slow)
It is always fun to put up the task managed on these to see them using 20-30MB of RAM with a large part of that being the current data base loaded in.
VLC & Blender are other examples of this.
It's kind of funny to imagine this parallel world where you send a PC of today back to the 70s. Whichever government got their hands on it would be keeping it ultra classified and hiding it away, like it was some device, too dangerous for the public, that could computationally solve any problem imaginable, create anything imaginable.
That's different from us devs losing efficiency from our deployment platform.
Functionality and graphics.
For instance 'dict.words' alone on Linux is 4.8MB. Arial Unicode is a 20MB-ish font. The icon for an application I work on is 400K. The Google Crashpad handler for handling crashes is somewhere around several MB.
A 4K true color display is 138 times larger than 640x480x16 colors.
With your examples, it could be:
- introduce global spell checker.
- have emoji?
- fix blurry icons?
- being able to search through crash logs?
- not having to switch between windows.
Do we need GBs instead of MBs for that? Why? Was that problem not fixed already? Could we not fix it in a way that didn't demand magnitudes more resources?
I'm asking, because I highly doubt that there's a technical reason that requires an improved piece of software or a solved problem,to require magnitudes more resources.
Sure, slack is far superior in UX to IRC. But could we really not get that UX without bloatware hogging my CPU, taking hundreds of MBs installation size and often biting off significant chunks of my memory? Is that truly, technically impossible?
A Word document isn't just text and some formatting sigils. Editing isn't just appending bytes to the end of a file descriptor.
It's a huge memory structure that has to hold an edit history so undo and track changes works, the spelling and grammar checker needs to live entirely in RAM since it runs in realtime as you edit, and the application itself has thousands of not millions of allocated objects for everything from UI elements to WordArt. The rendering engine needs to potentially hold a dozen fonts in memory for not just the system but any specified but not immediately used fonts from the base document template.
It's not like Google Docs is any lighter on RAM than Word. Features require memory. Fast features are usually going to require more memory.
People can use AbiWord if they want a much slimmer word processor. They could also just use nano and Markdown if they wanted even slimmer. But a lot of people want document sharing over the Internet with track changes, grammar checking, and the ability to drag in and edit an Excel spreadsheet.
The features used in native software follow a bathtub curve. But not just one but several. No two groups necessarily use the same sets of advanced/uncommon features.
> Functionality and graphics.
And massive amounts of telemetry.
But now functionality is moving to the cloud, we'll just be stuck with gigabytes for graphics and telemetry.
Having used that version of Word when it was the latest one, I can say the current ones have quite some added functionality (lots of very tiny things, and a few bigger ones), but I'm totally sure the same could be done with 10x less memory usage if MS would care about it. But there's no incentive to do it. Computers are faster and have lots of memory, and we don't depend on floppy disks any more. It would just cost them more money. Not saying that this is a good thing (I think the opposite, especially that I start thinking that software bloat might have a non-neglectable environmental impact), but as long as nobody complains strong enough (or as long as the EU doesn't come out with an anti-softare-bloat law... seems just a dream but who knows), that wont change. And I can clearly remember I had the same bloat-software feeling when I tried Office 2000 or XP, compared to Office 97, so there's nothing so new here.
As a final note, I've recently seen the source code of MS Word for Windows 1.0 on a GitHub repo (MS released it, see original release on Computer History Museum: https://computerhistory.org/blog/microsoft-word-for-windows-...). It was pure C, with even very large parts of code written in assembly! But the code is really ugly... totally incomparable to current C or C++ coding standards, patterns and language capabilities.
I saw it described once that software is like a gas - it expanded to take up the space we now have
You see it with live distros too. They used to be 700MB to fit on a CD-R, but now it's getting rare to find one that'll fit on a 2GB USB; although yay for 'minimal' gaining ground
Prime example of "Hardware is cheap and inefficiencies be damned"
It only takes longer to use while finding what I want among the bloated set of other things added.
Hah! Good one. It's unfortunate, but true.
[0] https://brave.com/privacy-updates/21-blocking-cookie-notices...
That seems to be contrary to the general opinion (at least on HN): google has become utterly irrelevant, serving mostly content farm AI generated junk type of blogspam, and google is more concerned about ad revenue than anything else (including results quality)
People here might say that Google has gotten terrible, but I would bet 99% of HNers still user Google dozens of times a day, just like everybody else in the world.
You’d owe the few extra cents.
You stay at 30% if annual proceeds continue to hit $1M/year. If not, you requalify for 15%.
Subscriptions are 15% for renewals (and maybe for all subs).
If your pulling in more than a few million a year from apple, and your not "gaming" or gaming the system I hear they are fairly open to negotiate. YMMV
When it comes to native apps, in the 2000s, this was the common attitude of users. But it's much harder to implement from a business perspective! Both in terms of business models, and in terms of dev time - having a bunch of possibly-incompatible versions lying around is a lot of overhead.
On the web, where most technical users understood this is technically impossible, they were willing to allow businesses to act differently, keep the software always-updated, and charge per usage. And since that's much easier and more lucrative for companies, they all switched to that.
(Now everyone kind of accepts that model, which is why today's Photoshop works via subscription, but the "damage" was done and the web won.)
Devs absolutely do not enjoy backporting bug fixes to 5 different LTS versions of their software and then getting user complaints because there's inevitably an important customer who is six versions back. It's inefficient with expensive dev time and it's better for the business to use that time to create new features.
edanm is correct, a lot of this is historical caused by very loud and angry tech users around the turn of the millennium. Want to know why Chrome won? When telling that story people tend to focus on performance or security, but that's not really it. Chrome won because Larry Page overrode all the internal screaming about silent web-style auto update for desktop apps. Oh boy, a whole lot of people really hated that idea, in fact Google had to develop their own software update engine from scratch to make it happen. Page didn't care. He understood that the ability to release a new version of web apps every week without the user noticing was a huge competitive advantage for the web, IE also updated in the background as part of the OS, and he wanted Google's desktop apps to have that same advantage.
Meanwhile Firefox stuck with the old model of rare releases and letting users choose whether to upgrade or not. It was a disaster. Old Firefoxes constantly annoyed web devs by preventing them from using new features. Security patches got reverse engineered and exploited. Still, Firefox's passionate fanbase loudly rejected the Chrome approach because they felt it took away their control.
Eventually the Mozilla guys accepted that they were wrong, their fans were wrong and Larry Page was correct. But it took years and in that time Chrome had built up a huge reputational advantage.
Businesses don't. Gotta find a way to sell the same shit every year.
Outside gaming, ai and big data, aka things for instance my parents don’t use at all, what limited feature wise? Browsers, sure, however my father prefers Teletext and newsgroups and Viditel (doesn’t exist anymore but he mentions it quite a lot) over ad infested slow as pudding websites. Email didn’t change since the 90s, word processors changed but not with stuff most people use (I still miss WP; it was just better imho; I went over to Latex because I find Word a complete horror show and that didn’t change), spreadsheets are used by pros and amateurs alike as a database mostly for making lists; nothing new there. You can go on and on; put an average user behind a 80s/90s pc (arguably after win95 release; DOS was an issue for many and 3.1 was horrible; or Mac OS) and they will barely notice the difference. Except for the above list of ai, big data, gaming and most importantly, browsers. Ai is mostly an api so that can be fixed (I saw a c64 openai chat somewhere) , big data is a very small % of humanity using that and gaming, well, depends what you like. I personally hate 3d games; I like 80s shmups and most people who game are on mobile playing cwazy diamonds or whatnot which I can implement on an msx 8 bit machine from the early 80s. Of course the massive multiplayer open world 3d stuff doesn’t work.
Anyway; as I said before here responding to what software/hardware to use for their parents; whenever someone asks me to revive their computer, I install Debian with i3 wm and dillo and ff as browser, Libreoffice and thunderbird. It takes a few hours to get used to but people (who are not in IT or any other computer fahig job) are flabbergasted by the speed, low latency and battery life. I did an x220 (with 9 cell) install last week; from win xp to the above setup; battery life jumped from 3 to 12 hours and everything is fast.
I install about 50 of those for people in my town throughout the year; people think they depend on certain software, but they really usually don’t. If they do, most things people ask for now work quite well under Wine. I have a simple script which starts an easy ‘Home Screen’ on i3 with massive buttons of their favourite apps which open on another screen (1 full screen per screen); people keep asking why Microsoft doesn’t do that instead of those annoying windows…
There are some useful resources: https://greycoder.com/a-list-of-text-only-new-sites/
There are also some tricks to have a lighter web browsing by default:
- try websites with netsurf, links or w3m first
- using a local web to gemini proxy to browse many websites with a lightweight gemini browser.
And you can go a long way by using an adblocker and/or disabling javascript by default using an extension with a toggle.
Everybody knew that updates weren’t free and they’d buy licenses when it was time to move on. But it’s hard to justify moving on when the software is still running fine even on Windows 10.
Human society has developed far slower throughout all history and prehistory, and that was OK. We’ve solved child mortality and we are doing just fine. But 1/3 of arable farmland is now desertified, insect populations are plummeting etc.
Urgency is needed the other way — in increasing cooperation. As we did ONE TIME with the Montreal Protocol and almost eliminated CFCs worldwide to repair the hole in the ozone layer
There are ways to optimize those things, though, which developers might not be bothering with anymore. The Design Patterns book used a word processor as the example when explaining the flyweight pattern for efficiently representing lots of objects. OLE objects like WordArt support different states[0] and don't necessarily have to be active at all times.
[0] https://learn.microsoft.com/en-us/windows/win32/com/object-s...
TO break even with apple you have about 80k a year all in cost to deal with all your refunds and charge backs.... after taxes, insurance and overhead that's 40-60k take home for a CS agent.
What is the charge back rate on digital goods? Im going to tell you that if your a small player it will be WAY higher than apple. Apple will cut a consumer off if they have a high refund rate, your CS agent will have no such insight.
%5-10 of your charges will just turn into refunds. Is that a process where you're killing license keys? Oh did you forget you now have infrastructure to run to issue and maintain said key? What is that going to cost you? Dont want to run like that... well ok then expect your return rate to go even higher. That discount CC processor is going to look at your refund and charge back rate and jack your fees up sky high (because that's the name of the game).
Once you get past a million bucks the open question is "do I do enough business to negotiate with apple". IN the case of a dry business oriented app, that has enough popularity to make that much, you might see apple willing to negotiate with you much sooner than a game dev who has sneaky buy options and huge charge back rates.
You can use chargeback protection on stripe or use a different payment provider which absorb the 15$ fee for chargebacks
But at $5 per user Apple is already much more expensive below the million threshold. It gets worse after a million, but it's already costing you tens of thousands before that. And again, you are comparing with one of the most expensive option on the market!
> after taxes, insurance and overhead that's 40-60k take home for a CS agent
Which, almost anywhere in the world, is more than you need to hire someone full time to work on your customer support! And no, what Apple provides is definitely not superior to a full time consumer support person.
The “value” that you pay for when dealing with Apple is access to their walled-off user base.
> the open question is "do I do enough business to negotiate with apple
This isn't an “open question”, it's a closed one: Apple isn't going to talk to you unless they think not giving you special treatment would get them antitrust issues. In your case or mine, it's not gonna happen.
But in a surprising number of cases, either customers don't have that choice (because the market hasn't provided a "fast" solution yet), or all things are not equal (say, the fast solution is fast because it's missing features that are crucial).
And this is why it always looks like customers are content with poorly-performing solutions.
I remember reading that even Google is worried about how crappy their results are becoming and at least some of the company sees it as a threat to itself.
But yes, you're right, Google is still very much mainstream and the #1 search engine. I'm just not certain it's because it's an amazing experience rather than "no other options".
(It might even be true! I'm just saying it's not new.)
Notarization is supposed to deter malware by a combination of static/dynamic analysis and attaching some real-world legal entity to any signed binary so law enforcement can follow up on if malicious activity is happening.
Analysis is not bulletproof and can be worked around.
The legal entity requirement is also trivial to nullify. At least in the UK, the company registration authority charges a nominal fee (payable by credit card - stolen if necessary) and puts you on the company register. Dun & Bradstreet scrapes that and that's how you get the DUNS number necessary to register for an Apple dev account. All of this is trivial to get through if you don't mind breaking the law and making up a few fake documents and providing a stolen CC (and assuming you're already planning to break the law by distributing malware, this is not a problem).
Finally, even if the "legal entity" bit was bulletproof, law enforcement just doesn't give a shit about the vast majority of online crime anyway.
All of these requirements are just a way to lock down access to the walled garden and put as many roadblocks to laymen trying to make their own software (in favor of big corps) masquerading as security theatre.
Firstly, stolen CCs tend to get reported especially if you make a big purchase. If you use a stolen CC to buy a developer certificate then it's going to get revoked the moment the real owner notices, and then your apps will be killed remotely by Apple before they've even been detected as malicious.
Still, the big win of notarization is that Apple can track down variants of your malware once it's identified and take them all out simultaneously. They keep copies of every program running on a Mac, so they can do clustering analysis server side. On Windows there's no equivalent of notarization, but the same task is necessary because otherwise malware authors can just spin endless minor variants that escape hash based detection, so virus scanners have to try and heuristically identify variants client side. This is not only a horrific resource burn but also requires the signatures to be pushed out to the clients where malware authors can observe them and immediately figure out how they're being spotted. Notarization is a far more effective approach. It's like the shift from Thunderbird doing spam filtering all on its own using hard-coded rules, to Gmail style server side spam filtering.
> All of these requirements are just a way to lock down access to the walled garden
I've been hearing this for over a decade now. In the beginning I believed it, but it's been a long time and Apple have never made macOS a walled garden like iOS is. There's no sign they're going to do it either. After all, at least some people have to be able to write new apps!
> analysis server side.
Are you sure about this ? I did not give apple permission to keep a copy of my software that I am writing.
I might be wrong here as I have been focused pretty much only on mobile, so feel free to correct me.
Switching ecosystems is nowhere near that trivial.
Ecosystem choices are dependent on content and tool investments, other devices owned, product groups, integrated technologies, network effects between people, between companies, customer relationships, existing phone payments, existing ecosystem familiarity and skills, on and on.
As for developers, they often need to be on the top 2-3 platforms to be a serious choice for customers.
Nothing wrong with highlighting different pros and cons of different ecosystems.
But a suggestion to switch ecosystems, without a very deep understanding of someone's particular situation, just isn't helpful advice.
With all that considered, I believe the extent of pushback that is possible is quite limited as long as the app technically works, but this is far from an accurate indicator of user happiness.
They're ok with it because they don't know it could be better. A spinner every 2 minutes? 12 minutes to open slack? They accept it as fact of life until a better software comes in, and now they're wondering why they haven't come across something like this sooner.
Only direct product properties should drive users' choices, everything else just raises the market entry barrier for potential competitors.
Would you? Because I would argue that CC processing is the point where you NEED near real time CS. Before that handling customer issues can be done better through forums, and you're going to get a lot of self service support from those.
>> (OP is talking about word doc apps, so we’re obviously not talking about indie $2 side project apps)
Your competing with free, libra office, Zoho writer (shockingly popular)... I would not know how to price the product to compete... 2 bucks a month as a trial? Would I pay 10 bucks a year if you were great? IF you got said productivity app past 100k users, getting to a million isnt a stretch (you have velocity and popularity).
Unless your doing something really slimy, your going to be able to get a better rate out of apple if you ask your rep.
Everybody pays for stuff online
The only Java desktop app I've ever used (on any platform) without frustration was Slay the Spire, and it only passes because it's a game and doesn't require desktop integration of any kind.
I use JDownloader sometimes, it's totally fine. Weka is bad, but not worse than other academic apps.
My senior developer mentors ended up having to effectively rewrite all of it because while it was technically correct and efficient, it broke all sorts of other good practices (eg didn't fit the existing coding style), or added in additional library dependencies without much thought towards long term maintainability and backwards compatibility.
It was taking so much time for the handful of already busy developers to go through my work that I had to learn to slow down, properly study the existing code and think about writing high quality code that fits the existing codebase. They didn't have the time to put down all their other work just to spend hours walking me through improving.
As you mention, it was like with learning art, it's impractical for a teacher to walk you through everything, you have to learn to identify errors and things you need to improve through your own meticulous study, relying on the teacher to give you hints when you're stuck.
A rejection of performance and compatibility as the core principles of software engineering in favor of “syntactic sugar” and “idiomatic Haskell”
In my 13 years in the industry, I’ve never worked at a place that valued that. More features faster, how many points this sprint is all that mattered. It’s put me off software engineering altogether.
Aesthetics matters.
Older M1 devices which are still very fast are available for much cheaper.
Bloat ware everywhere, especially browsers.
And that was on an IDE HDD, with memory speed, processor speed and cores a fraction of today, and 512MB of graphics memory or less.
Others in the thread talking about the heyday of older spreadsheet and document programs that were just as fast. So? I bet you could write a book on the new features and more advanced tools that MS Excel offers today compared to 1995.
We went from things taking minutes to taking seconds. So you could improve things by 50% and that could be VERY noticeable. (1min to 30s, for example.) If your app already launches in 500ms, 250ms is not going to make your laptop feel 2x faster even if it is. On top of that, since speed has been good enough for general computing for several years now, new laptops focus more on energy efficiency. I bet that new laptop has meaningfully better battery and thermal performance!
I'm sure you could, but it would be of interest to a relatively small audience. Excel 95 would be fine for about 90% of Excel users.
New expensive laptop had the same "fast" feeling which fade with new iterations of software. Browser takes insane amount of CPU and memory but isn't faster.
Maybe some intense CPU tasks like zipping folder is faster then ever, but I'm not zipping all day. However Slack is behaving like there is server side remote rendering for each screen...
That’s just your opinion though isn’t it
My opinion is my best friend. It is my life. I must master it as I must master my life.
Without me, my opinion is useless. Without my opinion, I am useless."
Using a shared object/DLL is the traditional way of doing so, but you can also accomplish this by providing the object files for your application to allow users to link their own substitutions statically.
The FSF explicitly permits this as documented here:
https://www.gnu.org/licenses/gpl-faq.en.html#LGPLStaticVsDyn...
You just have to open your source, that part which depends on Qt. It's not a real problem. But get a commercial license anyway, the cost is small compared to the other costs of developing your program, and you want to be friends with them.
(There's someone on HN who lives on a single-line modification of an open source program. Trust me, source availability of the source code of your client app won't really make a difference.)
If you're shipping overseas, you can probably ignore foreign taxes if you don't have a business nexus there. Especially if you have no desire to ever visit those countries. Basically just leave it up to your customers to pay whatever tax they owe.
[0]: https://en.wikipedia.org/wiki/South_Dakota_v._Wayfair,_Inc.
https://stripe.com/guides/introduction-to-eu-vat-and-vat-oss
Obviously, its not a good idea to bet your business on the courts not enforcing an EU fine when you can just add the VAT and cost of the handling hassle to the price for EU customers.
https://travel.state.gov/content/travel/en/legal/travel-lega...
This paperwork is what I believe a marketplace like the App Store or Amazon do for you under their own entity that you have to do yourself if you bypass their stores.
Please correct me if I'm wrong.
https://developer.apple.com/support/terms/apple-developer-pr...
Section 5.3: "By uploading Your Application to Apple for this digital notary service, You agree that Apple may perform such security checks on Your Application for purposes of detecting malware or other harmful or suspicious code or components, and You agree that Apple may retain and use Your Application for subsequent security checks for the same purposes."
Which leaves the certificate thing and while it’s an annoyance, it’s also nice as a software user to know that a program I’m running is the program it claims to be without much friction on my part, and the cost can’t be that prohibitive since I don’t remember the last time I ended up with an unsigned binary on my Mac, even for free software like TeX and friends or Aquamacs.
Ok, so your app tastes aren't that varied then (or maybe it's the memory), plenty of devs of various little utilities don't bother paying
Sometimes devs are locked into the tools they use. This situation is shit, but not uncommon.
Not true. The technical term for "ultimately need money" is discounted future cash flow. It is impossible to know for sure what price you have to charge for any particular item at any given time in order to optimise for this metric.
Realistically, the answer depends on the state of competition between platforms. We all know what that state is.
so it is true, you can't provide any top-level logic to justify 0, you need some facts
They want to be feudal lords, requiring us to pay a tithing for the privilege of selling something to customers.
You, of course, want to be their feudal lord and get access to all their customers by right while also requiring them to pay a tithing of their hardware sales to you since you advance "their ultimate benefit" (they wouldn't sell any hardware without software)
Yes obviously if you control the whole stack then you don't really need containers. If you're distributing software that is intended to run on Linux and not RHEL/Ubuntu/whatever then you can't rely on the userspace or packaging formats, so that's when people go to containers.
And of course if part of your infrastructure is on containers, then there's value in consistency, so people go all the way. It introduces a lot of other problems but you can see why it happens.
Back in around 2005 I wasted a few years of my youth trying to get the Linux community on-board with multi-distro thinking and unified software installation formats. It was called autopackage and developers liked it. It wasn't the same as Docker, it did focus on trying to reuse dependencies from the base system because static linking was badly supported and the kernel didn't have the necessary features to do containers properly back then. Distro makers hated it though, and back then the Linux community was way more ideological than it is today. Most desktops ran Windows, MacOS was a weird upstart thing with a nice GUI that nobody used and nobody was going to use, most servers ran big iron UNIX still. The community was mostly made up of true believers who had convinced themselves (wrongly) that the way the Linux distro landscape had evolved was a competitive advantage and would lead to inevitable victory for GNU style freedom. I tried to convince them that nobody wanted to target Debian or Red Hat, they wanted to target Linux, but people just told me static linking was evil, Linux was just a kernel and I was an idiot.
Yeah, well, funny how that worked out. Now most software ships upstream, targets Linux-the-kernel and just ships a whole "statically linked" app-specific distro with itself. And nobody really cares anymore. The community became dominated by people who don't care about Linux, it's just a substrate and they just want their stuff to work, so they standardized on Docker. The fight went out of the true believers who pushed against such trends.
This is a common pattern when people complain about egregious waste in computing. Look closely and you'll find the waste often has a sort of ideological basis to it. Some powerful group of people became subsidized so they could remain committed to a set of technical ideas regardless of the needs of the user base. Eventually people find a way to hack around them, but in an uncoordinated, undesigned and mostly unfunded fashion. The result is a very MVP set of technologies.
The dumpster fire at the bottom of that is libc and the C ABI. Practically everything is built around the assumption that software will be distributed as source code and configured and recompiled on the target machine because ABI compatibility and laying out the filesystem so that .so's could even be found in the right spot was too hard.
A few more:
* Seamless internationalization. If you're a native English speaker you probably never experienced the "fun" of dealing with French and Russian in the same text document. Pre-Unicode supported English + one other language, if that other language wasn't too weird.
* Lots of tiny life quality improvements. Eg, not seeing windows repaint costs a LOT of memory. Every window is present in RAM even if not being looked at so that when you switch to it you never see it paint.
* Stability. Windows 9x tried to be frugal by keeping a copy of everything in system32. That was called "DLL hell". So the current standard is that the app just packages every framework, so you may have a half dozen copies of Qt easily.
> Do we need GBs instead of MBs for that? Why?
Well, let me look at my AppImage:
3.8 GB total.
2.3 GB of dependencies. 2.1 GB is libnode, 128 MB is Qt Webengine.
1.4 GB application. 126 MB of JavaScript and UI images. The rest is mostly code.
For some programs, that hasn't changed. I use OneNote heavily to write some sort of personal info database I always look up when I forget something or need to reproduce a command verbatim quickly. The act of writing it and organizing the data also heavily reinforces my ability to memorize thing in my mind in and of itself too. So I'm quite fond of that little program.
When I tried to use it while learning Chinese I ended up having to turn off the spelling/grammar correction. It just can't function with two languages in the same notebook. All the Chinese text had the red squiggly lines warning you of a mistake and I found no way to enable the support for more than one language. You must select /one/ language for the spell checker in that program.
Or disable the spellchecker, which is what I did in the end.
I see that a lot with JavaScript apps. When they replace native, they often fail in details. Where e.g. my native text areas can handle multiple languages when spell checking. But where that diy or spellchuck.js npm version cannot.
But that the "cost" to do so, isn't what's technically required to improve it. You can achieve all these improvements, solve all these problems, probably without much more resource usage. Or negligible added resource usage.
Therefore my conclusion is that the reason e.g. slack usage x1000 what my old IRC or Jabber cliënt used, isn't technical. It's a deliberate choice made for reasons of budget, time to market or another trade-off.
I'm certain that Slack could build a client that does all what slack does, in a client that's hundred(s) of times snappier, smaller, less CPU and memory using. But probably not with their current pace, budget, team or wages.
That was never not the case.
Jabber makes a big usage of XML, which back in the day was very much seen as overkill. It requires a pretty complex parser, and increases the amount of data considerably.
They could have gone with a much more compact binary protocol with ID/length/value pairs, where there's not even field names, but say, a 16 bit integer where IDs are allocated from a central registry.
Even going back to DOS, you could shrink a program with measures like outputting "Error #5" instead of "File not found", and require the user to look up the code in a manual.
The newest macOS still needs more memory and suffer bloat but 8GB is still perfectly useable if you avoid google chrome. 8GB is also perfectly usable for Linux too.
For example, if disk space is abundant and very cheap, and optimizing software to use as little as disk space as possible is relatively more expensive than throwing more disk space at the problem, you shouldn't be surprised that software starts using more disk space than necessary, because what's being optimized is software development costs.
The "C ABI" and libc are a rather stable part of Linux. Changing the behaviour of system calls ? Linus himself will be after you. And libc interfaces, to the largest part, "are" UNIX - it's what IEEE1003.1 defines. While Linux' glibc extends that, it doesn't break it. That's not the least what symbol revisions are for, and glibc is a huge user of those. So that ... things don't break.
Now "all else on top" ... how ELF works (to some definition of "works"), the fact stuff like Gnome/Gtk love to make each rev incompatible to the prev, that "higher" Linux standards (LSB) don't care that much about backwards compat, true.
That, though, isn't the fault of either the "C ABI" or libc.
Good platforms allow you to build on newer versions whilst targeting older versions. Developers often run newer platform releases than their users, because they want to develop software that optionally uses newer features, because they're power users who like to upgrade, they need toolchain fixes or security patches or many other reasons. So devs need a "--release 12" type flag that lets them say, compile my software so it can run on platform release 12 and verify it will run.
On any platform designed by people who know what they're doing (literally all of the others) this is possible and easy. On Linux it is nearly impossible because the entire user land just does not care about supporting this feature. You can, technically, force the GNU ld to pick a symbol version that isn't the latest, but:
• How to do this is documented only in the middle of a dusty ld manual nobody has ever read.
• It has to be done on a per symbol basis. You can't just say "target glibc 2.25"
• What versions exist for each symbol isn't documented. You have to discover that using nm.
• What changes happened between each symbol isn't documented, not even in the glibc source code. The header, for example, may in theory no longer match older versions of the symbols (although in practice they usually do).
• What versions of glibc are used by each version of each distribution, isn't documented.
• Weak linking barely works on Linux, it can only be done at the level of whole libraries whereas what you need is symbol level weak linking. Note that Darwin gets this right.
And then it used to be that the problems would repeat at higher levels of the stack, e.g. compiling against the headers for newer versions of GTK2 would helpfully give your binary silent dependencies on new versions of the library, even if you thought you didn't use any features from it. Of course everyone gave up on desktop Linux long ago so that hardly matters now. The only parts of the Linux userland that still matter are the C library and a few other low level libs like OpenSSL (sometimes, depending on your language). Even those are going away. A lot of apps now are being statically linked against muslc. Go apps make syscalls directly. Increasingly the only API that matters is the Linux syscall API: it's stable in practice and not only in theory, and it's designed to let you fail gracefully if you try to use new features on an old kernel.
The result is this kind of disconnect: people say "the user land is unstable, I can't make it work" and then people who have presumably never tried to distribute software to Linux users themselves step in to say, well technically it does work. No, it has never worked, not well enough for people to trust it.
[1] Here's a guide to writing shared libraries for Linux that I wrote in 2004: https://plan99.net/~mike/writing-shared-libraries.html which apparently some people still use!
[2] Here's a script that used to help people compile binaries that worked on older GNU userspaces: https://github.com/DeaDBeeF-Player/apbuild
Yup, and I vendor a good number dependencies and distribute source for this reason. That and because distributing libs via package managers kinda stinks too, it's a lot of work. Id rather my users just download a tarball from my website and build everything local.
It's clearly an awful "patch" to outdated concepts on how commerce works compared to pre-internet, but it's what we have right now.
Right. Plus it might hinder your ability to travel freely in those jurisdictions as well which I'd like to avoid.
Disclosure: I am a co-founder of Galvix.
If someone buys an iPhone, Apple does not have the right to interpose themselves between that person and what they want to do with the iphone they bought. They have no right to a cut of the sales any more than the power company that provided the electrons to charge the battery.
What I want is for Apple to get out of the way.
> while also requiring them to pay a tithing of their hardware sales to you since you advance
What I want them to do, for a start, is to the same thing on ios that they already do on macos. I can already write a piece of software an sell it without forking 30% over to Apple.
The current situation where they feel entitled to a cut of every software sale that happens on ios, and veto power over it, is a wet dream that even Microsoft in the 90s wouldn't have thought they'd get away with.
Now I want to know more about this :-)
It's providing backwards compatibility (by symbol versioning). And that way allows for behaviour to evolve while retaining it for those who need that.
I would agree it's possibly messy. Especially if you're not willing or able to change your code providing builds for newer distros. That said though... ship the old builds. If they need it only libc, they'll be fine.
(the "dumpster fire" is really higher up the chain)
Your solution solves a made up problem that nobody cares about, and doesn't solve the one that actually matters, which is to successfully make and distribute good software to users.
Someone shouldn't have to add the the fine print line "Assume I am talking about things that matter, instead of things that don't" to every statement or opinion that they have.
A lot of it introduced from 2017 onwards and I think now it says something akin to "this application will hack your computer and is a virus" and you need to click the smaller hidden "ignore"s a few times to do what you want.
well as someone who runs a few unsigned binaries myself. Its not hard if you know what to do but apple makes a big deal about how its "unsafe" and this freaks non tech people out.
https://web.archive.org/web/20060925013545/http://www.making...
As this is such a pointlessly contrived interpretation of the term "logic" in this context, I chose to use a different one: Is there a set of empirical circumstances under which an optimisation algorithm could conclude that the optimal price is zero? The answer to that is clearly yes.
Now, what exactly is the point of you insisting on your wrong interpretation?
Here are the most commonly used options:
- Go LGPL. Sure, you will need to ship binaries and libs, but there are tools within the SDK that do this automatically for you (windeployqt, macdeployqt, etc.). And as others have stated, it is a problem that was solved years ago.
- Go Commercial to link statically. If you are a single developer, there is an annual license available for $499 (up to $100k yearly revenue).
It might not be worth the price, but that is hardly ridiculous. It is quite believable to get a 4% productivity improvement from appropriate tooling. You need to do a cost-benefit analysis to determine the answer to that question.
https://www.youtube.com/watch?v=zvfD5rnkTws
Seriously, though, I've had the Windows Defender thing happen to freshly compiled binaries I made. The only way to prevent it from happening is to sign your binaries, or submit them individually to Microsoft using your Microsoft account for malware analysis.
It flagged the binary as being some sort of trojan (which name I looked up and found that it was a Windows Defender designation for "I don't know the provenance of this binary so I'm going to assume it's bad") and quarantined it.
This got an audible laugh out of me.
> Good platforms allow you to build on newer versions whilst targeting older versions.
I haven't been doing this for 20 years (13), but I've written a fair amount of C. This, among other things, is what made me start dabbling with zig.
~ gcc -o foo foo.c
~ du -sh foo
16K foo
~ readelf -sW foo | grep 'GLIBC' | sort -h
1: 0000000000000000 0 FUNC GLOBAL DEFAULT UND __libc_start_main@GLIBC_2.34 (2)
3: 0000000000000000 0 FUNC GLOBAL DEFAULT UND puts@GLIBC_2.2.5 (3)
6: 0000000000000000 0 FUNC GLOBAL DEFAULT UND __libc_start_main@GLIBC_2.34
6: 0000000000000000 0 FUNC WEAK DEFAULT UND __cxa_finalize@GLIBC_2.2.5 (3)
9: 0000000000000000 0 FUNC GLOBAL DEFAULT UND puts@GLIBC_2.2.5
22: 0000000000000000 0 FUNC WEAK DEFAULT UND __cxa_finalize@GLIBC_2.2.5
~ ldd foo
linux-vdso.so.1 (0x00007ffc1cbac000)
libc.so.6 => /usr/lib/libc.so.6 (0x00007f9c3a849000)
/lib64/ld-linux-x86-64.so.2 => /usr/lib64/ld-linux-x86-64.so.2 (0x00007f9c3aa72000)
~ zig cc -target x86_64-linux-gnu.2.5 foo.c -o foo
~ du -sh foo
8.0K foo
~ readelf -sW foo | grep 'GLIBC' | sort -h
1: 0000000000000000 0 FUNC GLOBAL DEFAULT UND __libc_start_main@GLIBC_2.2.5 (2)
3: 0000000000000000 0 FUNC GLOBAL DEFAULT UND printf@GLIBC_2.2.5 (2)
~ ldd foo
linux-vdso.so.1 (0x00007ffde2a76000)
libc.so.6 => /usr/lib/libc.so.6 (0x0000718e94965000)
/lib64/ld-linux-x86-64.so.2 => /usr/lib64/ld-linux-x86-64.so.2 (0x0000718e94b89000)
(Haven't tested what it would link against where that not given)
I'd only like to add one thing here ... on static linking.
It's not a panacea. For non-local applications (network services), it may isolate you from compatibility issues, but only to a degree.
First, there are Linux syscalls with "version featuritis" - and by design. Meaning kernel 4.x may support a different feature set for the given syscall than 5.x or 6.x. Nothing wrong with feature flags at all ... but a complication nonetheless. Dynamic linking against libc may take advantage of newer features of the host platform whereas the statically linked binary may need recompilation.
Second, certain "features" of UNIX are not implemented by the kernel. The biggest one there is "everything names" - whether hostnames/DNS, users/groups, named services ... all that infra has "defined" UNIX interfaces (get...ent, get...name..., ...) yet the implementation is entirely userland. It's libc which ties this together - it makes sure that every app on a given host / in a given container gets the same name/ID mappings. This does not matter for networked applications which do not "have" (or "use") any host-local IDs, and whether the DNS lookup for that app and the rest of the system gives the same result is irrelevant if all-there-is is pid1 of the respective docker container / k8s pod. But it would affect applications that share host state. Heck, the kernel's NFS code _calls out to a userland helper_ for ID mapping because of this. Reimplement it from scratch ... and there is absolutely no way for your app and the system's view to be "identical". glibc's nss code is ... a true abyss.
Another such example is (another "historical" wart) timezones or localization. glibc abstracts this for you, but language runtime reimplementations exist (like the C++2x date libs) that may or may not use the same underlying state - and may or may not behave the same when statically compiled and the binary run on a different host.
Static linking "solves" compatibility issues also only to a degree.
Right, but I was able to do it as a whole. I didn't have to do it per symbol.
I think less than 1% of the users use it on mobile, but it's designed as a mobile interface.
To scroll you need to click and drag, or you need to click 5px buttons. Regular mouse scroll doesn't work.
If you're doing a GUI, you have no reason to be doing canvas manually.