7 ms·

This is such a lovely idea on paper, but how do you put that into practice? Is the problem the monolithic design of a web browser, which now is basically an ope

This is such a lovely idea on paper, but how do you put that into practice? Is the problem the monolithic design of a web browser, which now is basically an operating system all its own? Is the solution to break that into its component parts, so that each part can be maintained by a smaller group, and composed together to produce the browser as a whole?

stirner 9y ago
The problem is the complexity of all the specifications that make up the Web. These specifications are heavily influenced by whatever major browsers already are doing (i.e. influenced by a handful of huge tech companies). These companies have a vested interest in making it hard to make a new, competitive browser. I think the Web is unredeemable at this point; there is so much entrenched complexity, ugly hacks, centralization, and misuse of various technologies that it can never be undone. The only solution is to refuse to contribute to the Web at all, which is hardly an option for most of us here.
- thomastjeffery 9y ago
  The solution is to fork, or start over. I like the latter idea.
  - hossbeast 9y ago
    Are there any ongoing efforts in that regard? I guess IPFS, but even then, you're serving html.
    - xj9 9y ago
      its slow going, but that's what i'm trying to do with heropunch[0]. our goal is to create a p2p application platform using a handful of libre technologies: secure scuttlebutt, ipfs, lxc, enlightenment foundation libraries, rust, and elixir. [1]: https://www.heropunch.io https://www.heropunch.io
  - krapp 9y ago
    Great. I support this. Go build your new web and leave those of us who believe the existing web, despite its faults, has value, be.
    - abiox 9y ago
      just about anything people invest time into is going to have some sort of value, 'despite faults'. that doesn't really speak to the value of progress, reflecting on the status quo, or creating something fresh with lessons learned from the past. i don't really get the 'leave you be' bit.
      - krapp 9y ago
        >i don't really get the 'leave you be' bit. Simple. More than a few people here seem to believe, as the grandparent comment suggests, that the web is a lost cause, or irredeemable, and all those people seem to want to do is to constantly complain about it. I'm simply suggesting that if people feel the web has nothing to offer them, that it would be more productive for them to kindly leave it for a network that better suits their needs. Otherwise, rather than wanting to "fork the web and start over" they could consider working to improve what we have.
        abiox 9y ago
        viewing the web as a lost cause or irredeemable is not the same as saying it has nothing to offer. 'leaving the web' also doesn't seem very pragmatic, at least until something better is available.
        krapp 9y ago
        >viewing the web as a lost cause or irredeemable is not the same as saying it has nothing to offer I disagree - if it has something to offer then it's not a lost cause. A lost cause by definition isn't worth saving, or even engaging with. > 'leaving the web' also doesn't seem very pragmatic, at least until something better is available. The comment I replied to earlier suggested that the only reasonable solution to the web was to fork it or start over, with starting over being preferable. I'm merely suggesting that someone should actually get started on that. Or maybe revive Gopher. I hear that's still around.
- Retric 9y ago
  I think we could build a simple standard that does the useful parts of layout / CSS and includes a sane scripting language in a reasonable timeframe. It would need to be accessible both for users and developers. The hard part would be enforcing behavior so you don't end up with venders adding their own bits which destroy the entire point of the thing.
- sliverstorm 9y ago
  I think the Web is unredeemable at this point; there is so much entrenched complexity, ugly hacks, centralization, and misuse of various technologies that it can never be undone. Somewhere, an AMP developer coughs indignantly
rjbwork 9y ago
I think this idea has some merit to it, but I don't have any experience in building that kind of software. This is essentially how teams that build huge software systems work. Like for instance, operating systems like windows/linux. The various teams at Microsoft, or trusted comitters for Linux, organize their various modules, components, subsystems, etc. independently, and eventually compose them into a coherent functioning whole to ship the whole system.
- thomastjeffery 9y ago
  The biggest difficulty is tying components together. There are a few methods: 1. Libraries linked by binaries. This is the usual method, but it generally demands modules share some things, which usually couples them too tightly to their implementation. 2. IPC (Inter-Process Communication) There are a few ways to accomplish this. Some are OS specific (named pipes), others are fairly generic (sockets). This requires modules to share a language, and has some overhead, but at least they aren't coupled. 3. Microkernel Essentially the best points of 1 and 2, but generally a design challenge itself. The biggest advantage to modular design is that you don't need to create all the modules before you get something useful.
thomastjeffery 9y ago
Design and create the equivalent of a web browser in a modular fashion. Yes, that is a lofty goal. That does not make it unreasonable. > Is the solution to break that into its component parts, so that each part can be maintained by a smaller group, and composed together to produce the browser as a whole? Possibly, but currently, a "web" browser depends on its monolithic qualities. A web page is a DOM, defined by HTML, styled by CSS, and manipulated by JavaScript. I think it may be time to start designing something less inherently monolithic. One advantage to modularity is that we don't need to finish before we can use it.
- xj9 9y ago
  we are slowly working towards this goal[0], we try to team up with other projects and develop this stuff cooperatively. our focus is on the user and developer experience side of things. creating good comfy tools for developers and fighting for very high quality UX on the user side. [0]: https://heropunch.io https://heropunch.io
- snaky 9y ago
  That's very nontrivial task, if feasible at all, considering current overcomplicated state of affairs in CSS, HTML, JS, graphical, font-rendering and networking fields of browser's interest. > Guys, there is a _reason_ why microkernels suck. This is an example of how things are _not_ "independent". The filesystems depend on the VM, and the VM depends on the filesystem. You can't just split them up as if they were two separate things (or rather: you _can_ split them up, but they still very much need to know about each other in very intimate ways). http://yarchive.net/comp/linux/user_space_filesystems.html http://yarchive.net/comp/linux/user_space_filesystems.html
  - thomastjeffery 9y ago
    > That's very nontrivial task I imagine so. And?
VarFarYonder 9y ago
James Mickens gave a great talk about how we could build a better browser: https://www.youtube.com/watch?v=1uflg7LDmzI https://www.youtube.com/watch?v=1uflg7LDmzI
- AlexCoventry 9y ago
  You're right, this is great. Is the Atlantis source code available?
  - VarFarYonder 9y ago
    I've looked for it previously and not found anything. I'm hoping that MS are working on it in secret, but seems unlikely.
- herogreen 9y ago
  Amazing talk, thanks!

stirner 9y ago

The problem is the complexity of all the specifications that make up the Web. These specifications are heavily influenced by whatever major browsers already are doing (i.e. influenced by a handful of huge tech companies). These companies have a vested interest in making it hard to make a new, competitive browser. I think the Web is unredeemable at this point; there is so much entrenched complexity, ugly hacks, centralization, and misuse of various technologies that it can never be undone. The only solution is to refuse to contribute to the Web at all, which is hardly an option for most of us here.

thomastjeffery 9y ago
The solution is to fork, or start over. I like the latter idea.
- hossbeast 9y ago
  Are there any ongoing efforts in that regard? I guess IPFS, but even then, you're serving html.
  - xj9 9y ago
    its slow going, but that's what i'm trying to do with heropunch[0]. our goal is to create a p2p application platform using a handful of libre technologies: secure scuttlebutt, ipfs, lxc, enlightenment foundation libraries, rust, and elixir. [1]: https://www.heropunch.io https://www.heropunch.io
- krapp 9y ago
  Great. I support this. Go build your new web and leave those of us who believe the existing web, despite its faults, has value, be.
  - abiox 9y ago
    just about anything people invest time into is going to have some sort of value, 'despite faults'. that doesn't really speak to the value of progress, reflecting on the status quo, or creating something fresh with lessons learned from the past. i don't really get the 'leave you be' bit.
    - krapp 9y ago
      >i don't really get the 'leave you be' bit. Simple. More than a few people here seem to believe, as the grandparent comment suggests, that the web is a lost cause, or irredeemable, and all those people seem to want to do is to constantly complain about it. I'm simply suggesting that if people feel the web has nothing to offer them, that it would be more productive for them to kindly leave it for a network that better suits their needs. Otherwise, rather than wanting to "fork the web and start over" they could consider working to improve what we have.
      - abiox 9y ago
        viewing the web as a lost cause or irredeemable is not the same as saying it has nothing to offer. 'leaving the web' also doesn't seem very pragmatic, at least until something better is available.
        krapp 9y ago
        >viewing the web as a lost cause or irredeemable is not the same as saying it has nothing to offer I disagree - if it has something to offer then it's not a lost cause. A lost cause by definition isn't worth saving, or even engaging with. > 'leaving the web' also doesn't seem very pragmatic, at least until something better is available. The comment I replied to earlier suggested that the only reasonable solution to the web was to fork it or start over, with starting over being preferable. I'm merely suggesting that someone should actually get started on that. Or maybe revive Gopher. I hear that's still around.
Retric 9y ago
I think we could build a simple standard that does the useful parts of layout / CSS and includes a sane scripting language in a reasonable timeframe. It would need to be accessible both for users and developers. The hard part would be enforcing behavior so you don't end up with venders adding their own bits which destroy the entire point of the thing.
sliverstorm 9y ago
I think the Web is unredeemable at this point; there is so much entrenched complexity, ugly hacks, centralization, and misuse of various technologies that it can never be undone. Somewhere, an AMP developer coughs indignantly

rjbwork 9y ago

I think this idea has some merit to it, but I don't have any experience in building that kind of software. This is essentially how teams that build huge software systems work. Like for instance, operating systems like windows/linux. The various teams at Microsoft, or trusted comitters for Linux, organize their various modules, components, subsystems, etc. independently, and eventually compose them into a coherent functioning whole to ship the whole system.

thomastjeffery 9y ago
The biggest difficulty is tying components together. There are a few methods: 1. Libraries linked by binaries. This is the usual method, but it generally demands modules share some things, which usually couples them too tightly to their implementation. 2. IPC (Inter-Process Communication) There are a few ways to accomplish this. Some are OS specific (named pipes), others are fairly generic (sockets). This requires modules to share a language, and has some overhead, but at least they aren't coupled. 3. Microkernel Essentially the best points of 1 and 2, but generally a design challenge itself. The biggest advantage to modular design is that you don't need to create all the modules before you get something useful.

thomastjeffery 9y ago

Design and create the equivalent of a web browser in a modular fashion. Yes, that is a lofty goal. That does not make it unreasonable. > Is the solution to break that into its component parts, so that each part can be maintained by a smaller group, and composed together to produce the browser as a whole? Possibly, but currently, a "web" browser depends on its monolithic qualities. A web page is a DOM, defined by HTML, styled by CSS, and manipulated by JavaScript. I think it may be time to start designing something less inherently monolithic. One advantage to modularity is that we don't need to finish before we can use it.

xj9 9y ago
we are slowly working towards this goal[0], we try to team up with other projects and develop this stuff cooperatively. our focus is on the user and developer experience side of things. creating good comfy tools for developers and fighting for very high quality UX on the user side. [0]: https://heropunch.io https://heropunch.io
snaky 9y ago
That's very nontrivial task, if feasible at all, considering current overcomplicated state of affairs in CSS, HTML, JS, graphical, font-rendering and networking fields of browser's interest. > Guys, there is a _reason_ why microkernels suck. This is an example of how things are _not_ "independent". The filesystems depend on the VM, and the VM depends on the filesystem. You can't just split them up as if they were two separate things (or rather: you _can_ split them up, but they still very much need to know about each other in very intimate ways). http://yarchive.net/comp/linux/user_space_filesystems.html http://yarchive.net/comp/linux/user_space_filesystems.html
- thomastjeffery 9y ago
  > That's very nontrivial task I imagine so. And?

VarFarYonder 9y ago

James Mickens gave a great talk about how we could build a better browser: https://www.youtube.com/watch?v=1uflg7LDmzI https://www.youtube.com/watch?v=1uflg7LDmzI

AlexCoventry 9y ago
You're right, this is great. Is the Atlantis source code available?
- VarFarYonder 9y ago
  I've looked for it previously and not found anything. I'm hoping that MS are working on it in secret, but seems unlikely.
herogreen 9y ago
Amazing talk, thanks!