The author has very limited knowledge of the web, its history, or the W3C's ambitions (not to mention WHATWG, IETF, ECMA, etc.). His view of the web in general is bafflingly myopic, with an almost completely arbitrary understanding of the businesses discussed. It should come as no surprise that the conclusions he draws are non sequitur. So why has this been upvoted? Why is this on HN at all? It's ignorant nonsense. T…
When will web browsers be complete?
181–190 of 213 posts
Re: When will web browsers be complete?
#182Web standards still have some glaring gaps that I'm not sure the current approach can tame. I propose "the web" be split into three standards:
1. Media/Games/Art/Video
2. Documents (existing HTML may be good enough with minor tweeks)
3. Productivity: CRUD and Data oriented interactive GUI
By splitting, each standard can FOCUS on doing what it does best rather than the watered-down one-size-fits-all we have now. A given browser may support all 3, somewhat like how Java Applets and Flash could run inside of HTML documents. (They may be pluggins and/or independent browsers.)
But unlike Applets & Flash, the standards wouldn't try to be full virtual OS's. Dump as much on the server as you can to keep client standards lean, and reduce the need for scripting by including common behaviors.
For example, a common GUI idiom is for a button to activate a window or form. Rather than rely on scripting, that action could be built into the GUI markup:
Click to open My Window
Or maybe:
The point is that if common GUI idioms and behaviors are built into the GUI markup standard, scripting (typically JavaScript) is not needed as often. The GUI Markup Language would be state-friendly and natively interactive, unlike HTML.Re: When will web browsers be complete?
#183I have a complete unlovely thought; web browser needs a rewrite with thin web specs (imagine javascript the good part book meme). People think it's impossible to rewrite, but I believe it's only because it's a mess. Inventing a new programming language and new DOM api that works greatly with webassembly probably worth it, consider that the next 10 years the web is still not going anywhere. You may laugh at Flash .. b…
The problem is that so much of technology is standing on the shoulders of giants. If we re-write, how much of the giants do we decide to lop-off?
For example, could you get Google, Apple, Mozilla, to all agree where to draw the line? Should we still use HTTP? Should everything use QUIC? An entirely new standard? What about DNS? But let's say they do agree.
Then what, they create entirely new browsers (I'll call Chrome2, Safari2, etc). Maybe they don't even use the DOM or HTML at all, but an entirely new setup. A side note – if the web was re-built today it probably wouldn't have "View Source." As you said, I imagine it'd lean into WASM and "isomorphic" code, which could just be WebGL/canvas/some new paradigm/etc.
So then what... everyone says abandon my original website and download a new browser to access the new one? Because that's what throwing out backwards compatibility looks like. And that seems like a non-starter to most people.
Browsers are too big to fail; too big to be thrown out and re-written. Our hope is incremental, not revolutionary. Things like WASM in browsers let us pursue new paradigms while also being backwards compatible.
Lastly, I'd say the incentives for a new web would be worse than an old web. The original web was written by tinkerers and tech startups like Netscape/Mozilla. The new web would be written by multibillion dollar companies who like and know how to build moats. It would have microtransactions (not necessarily bad by itself) and I worry would look more like an app store.
Re: When will web browsers be complete?
#184Earlier quoted context omitted.
That 2007 passage really brought me to a grinding hault. Toddlers are using the web these days from the moment they can swipe a screen. I have a hard time believing anybody on GitHub over the age of 20 only began using the internet in 2007.
I didn't regularly use the Internet until I was 13ish. I wanted to be fair in the timeframe so that's why I said that. I probably encountered the Internet many many times before then, but it was not for regular usage.
The age of 13 seems to be a common thread with a lot of developers -- I know of 5 others who will point to that age as "when they started writing software". It was age 13 when I started assembling and selling PCs (in the 90s) and began exploring software development and before my 14th birthday that I had re-written a BBS application in PASCAL.
Like you, I had been messing around with my (horrible) computer[0] since it arrived in our house and I was presented with an `A>` with no clue how to proceed, but it wasn't until I was 13 that my Dad presented me with a plan and a book to fix that problem[1]. I wanted a 386 so I could play Wing Commander (and see more than the 4 colors that my CGA display allowed -- loved the LSD trip that games like Kings Quest would put you through while trying to render decent graphics using practically nothing, but I was ready to be done with that).
I remember thinking my dad was insane -- he owned a business selling things to large automotive manufacturers ... nothing event adjacent to computers. And a computer was easily the most complicated consumer electronics device a person could own. Having had to fix our 8088 a few times, and being generally the most "handy person" I've ever known, his response was a dismissive "they're all just a bunch of parts that you buy and put together". And while I'm pretty sure he helped me build/troubleshoot the first couple of them (we had some bad components early on from a vendor in California), I generally built/tested/assembled all of the few hundred that we sold, myself. If you want a 13-year-old to behave more like a "Grown Up", well, when your Dad's friends are paying you thousands of dollars to build them a computer, you start to act like someone who's worthy of receiving that kind of trust.
The best part: we built the first several computers to pay for the one we wanted for ourselves[2]. My dad owned his company. I'm willing to wager that he could have afforded to just buy the thing for me and probably wouldn't have felt all that guilty about spoiling his kid since I was using it 99% of the time to do things other than play games. But this served as a forcing function -- I got the exact computer with the exact specifications that I wanted without having to think about the cost of those things provided I could sell enough computers to purchase it. I did. And I got to buy a convertible for my first car with money I had earned.
I'm seeing it now, in my own kids, two of which turn 13 in the next few months. They're working with Roblox's developer tools making new worlds and I'm staring over their shoulder thinking: this application is easily as complicated to use as any of the design tools I've been exposed to and these kids are just ripping through it like it's "not work" (if only they were getting paid!).
I've been looking for things to motivate them to step out of the "game world" and realize that the things they're doing could be done outside of Roblox -- they could make real applications, but much like I didn't believe my dad when he handed me that book, they haven't had that "ah ha" moment, yet.
[0] My friends all had Commodore 64s and such that played really cool (if slowly loading) games and cost a few hundred dollars. We had a $3,000 IBM 8088 clone with a 10MB hard drive and floppy drive that ... wasn't fun.
[1] I can't remember the title completely but it was something like "Build an 80486 PC and Save A Bundle"
[2] My 80486, if memory serves, cost just shy of $10,000. We went with what would be called "server hardware" these days: SCSI over (very new) IDE or (very old) MFM -- we went SCSI with a full height 360MB drive (40MB drives were huge), a very nice video card and a 18" CRT display (14" was 'normal' with anything above starting at twice the price of the best 14" display). And two modems... I ran a multi-node BBS on custom software originally based on Telegard.
Re: When will web browsers be complete?
#185Earlier quoted context omitted.
On mobile blame goes to Apple for not allowing other engines. You could have used Firefox in iOS devices as well.
That is what saves the day of Safari actually being relevant at all, instead of Chrome overall. I am an happy Firefox user, but don't fool myself, without Safari, Web === ChromeOS.
Re: When will web browsers be complete?
#186When this essay started with "As someone who has used the Web since 2007" I immediately felt old, so thanks for that. As someone who has been using the Web since 1994 (and I feel like I was late to the game), I can assure the author their perspective has a lot of recency bias and this isn't something to worry about, in the general sense. Or worry about it, because it already happened a while ago. Take your pick. Loca…
> This essay could've been written in 2002...about Internet Explorer. I've been reading a whitepaper from ~1997 on Microsoft's Distributed COM (DCOM) last night, and as I was reading it, I couldn't stop thinking: this is microservices. And it was really more than that - at least on the surface, this was microservices + orchestration + autoscaling + serverless, as well as, via Internet Explorer, component-based UIs wi…
For some reason that I fail to understand, it seems that most computer protocols and most computer file formats are designed under the unstated assumption that those using them will not have access to a computer.
Re: When will web browsers be complete?
#187Earlier quoted context omitted.
I didn't regularly use the Internet until I was 13ish. I wanted to be fair in the timeframe so that's why I said that. I probably encountered the Internet many many times before then, but it was not for regular usage.
A little off topic, but thanks for sharing: The age of 13 seems to be a common thread with a lot of developers -- I know of 5 others who will point to that age as "when they started writing software". It was age 13 when I started assembling and selling PCs (in the 90s) and began exploring software development and before my 14th birthday that I had re-written a BBS application in PASCAL. Like you, I had been messing a…
Re: When will web browsers be complete?
#188Earlier quoted context omitted.
That is what saves the day of Safari actually being relevant at all, instead of Chrome overall. I am an happy Firefox user, but don't fool myself, without Safari, Web === ChromeOS.
Then on mobile it would be Safari. And mobile would be the future growth.
Re: When will web browsers be complete?
#189Earlier quoted context omitted.
You literally couldn't have used the web much earlier than 1994 unless you were running a NeXT machine. I know because I've been on the net since 1988 and 1994 was when I first started browsing the web. If you were on Windows it was tricky then as you had to download and configure your own tcp/ip stack. That was possible due to the ingenuity of Peter Tattam who sold software that let you do that. I met him back then…
You could. X.25. Gopher.
Re: When will web browsers be complete?
#190Earlier quoted context omitted.
I think we should separate ideas/proof of concepts that were in the field and presented as having potential, and implementations that are largely adopted and reliable. There will always be a lag between the two, and it won’t just be ‘reinventing the wheel’. I had a look at CORBA and it was horrible in many ways. Interestingly it’s something that works and is proven at large scale but nobody would want to have it for…
How on earth can anyone think that each application having to individually implement support for a bespoke "dark mode" is a better situation than having more granular controls in the OS and applications just follow the UI guidelines is beyond me. It's bullshit like that that makes me want to leave IT forever.
Yes, in terms of raw flexibility exposed to the user, customizable color schemes look great on the surface. They work to some degree, but fail utterly for a wide range of cases in practice. Not having them is progress after a fashion because it means giving up on an flawed idea. I will admit, though, that it is not easy to see why it is flawed when you haven't tried to design custom controls that have to convey more state than can be expressed using the standard OS colir palette.