3 ms·

Any estimates on what the economic cost of entry is for developing a competetive browser? As someone who has no background doing web development, would also li

by dman 9y ago

Any estimates on what the economic cost of entry is for developing a competetive browser? As someone who has no background doing web development, would also like to hear from web devs on what proportion of a project is spent building features vs handling edge cases in browsers (like the one that was hit here in the article).

Gaelan 9y ago
Blink wrapper or new engine? If you want chrome clones, they’re a dime a dozen.
- dman 9y ago
  New engine.
kbyatnal 9y ago
Well considering that Chrome/Firefox have more lines of code than the Mars Rover, Large Hadron Collider, Hubble Space Telescope, Linux Kernel, and US Military drones, I'd say the cost is pretty significant. http://www.informationisbeautiful.net/visualizations/million-lines-of-code/ http://www.informationisbeautiful.net/visualizations/million...
Sylos 9y ago
> Any estimates on what the economic cost of entry is for developing a competetive browser? I doubt you're gonna get numbers, but I can give you two rough reference points: 1) All popular modern browsers use browser engines that started out their lives in the last millennium: - Edge uses EdgeHTML, which was forked from IE's Trident, which was first released in 1997. - Firefox uses Gecko, which started development in 1997. - Chrome/Opera/Vivaldi use Blink, which was forked from WebKit (which Safari uses), which in turn was forked from KHTML, which started development somewhen around 1998. So, all of these had their architecture laid out before multi-core CPUs were a thing in desktop PCs and before the web became more than just text with the occasional image thrown in. As a result, all of them lack severely in parallelism and use of the GPU, and writing a browser engine from scratch would allow you to get these things right and give you ridiculously better performance than all other browser. But no one's done it yet, because of this insanely high cost to redevelop everything. Which brings us to reference point 2), Mozilla has actually started work on writing a browser engine from scratch in 2013. The project is called Servo and they even find it worthwhile to basically develop their own programming language, Rust, for writing Servo in it. And well, Servo shows off quite nicely the aforementioned potential in writing a browser engine from scratch, but it also still explodes spectacularly, trying to render out ACID3, which was released in 2008 to try to represent what a good browser should have been capable of back then. Now, Mozilla is not working on this full-pelt, but it is a big organisation with lots of browser-making know-how. They're also sort of bootstrapping Servo by including production-ready Servo components into Gecko, resulting in the maintenance work being shared. So, basically I wouldn't expect any new player to enter the market in at least the next ten years.
- steveklabnik 9y ago
  In my understanding, the Servo team has never focused on ACID 3, instead prioritizing features that are actually used by the top X websites.
- dman 9y ago
  Thanks for a detailed response!
itomato 9y ago
Read up on the history of Konqueror and WebKit to get an idea of what it takes on the render engine side. Marketing, extensibility, feature-parity are in another realm, and remain a moving target.

by dman 9y ago

Gaelan 9y ago

Blink wrapper or new engine? If you want chrome clones, they’re a dime a dozen.

dman 9y ago
New engine.

kbyatnal 9y ago

Well considering that Chrome/Firefox have more lines of code than the Mars Rover, Large Hadron Collider, Hubble Space Telescope, Linux Kernel, and US Military drones, I'd say the cost is pretty significant. http://www.informationisbeautiful.net/visualizations/million-lines-of-code/ http://www.informationisbeautiful.net/visualizations/million...

Sylos 9y ago

> Any estimates on what the economic cost of entry is for developing a competetive browser? I doubt you're gonna get numbers, but I can give you two rough reference points: 1) All popular modern browsers use browser engines that started out their lives in the last millennium: - Edge uses EdgeHTML, which was forked from IE's Trident, which was first released in 1997. - Firefox uses Gecko, which started development in 1997. - Chrome/Opera/Vivaldi use Blink, which was forked from WebKit (which Safari uses), which in turn was forked from KHTML, which started development somewhen around 1998. So, all of these had their architecture laid out before multi-core CPUs were a thing in desktop PCs and before the web became more than just text with the occasional image thrown in. As a result, all of them lack severely in parallelism and use of the GPU, and writing a browser engine from scratch would allow you to get these things right and give you ridiculously better performance than all other browser. But no one's done it yet, because of this insanely high cost to redevelop everything. Which brings us to reference point 2), Mozilla has actually started work on writing a browser engine from scratch in 2013. The project is called Servo and they even find it worthwhile to basically develop their own programming language, Rust, for writing Servo in it. And well, Servo shows off quite nicely the aforementioned potential in writing a browser engine from scratch, but it also still explodes spectacularly, trying to render out ACID3, which was released in 2008 to try to represent what a good browser should have been capable of back then. Now, Mozilla is not working on this full-pelt, but it is a big organisation with lots of browser-making know-how. They're also sort of bootstrapping Servo by including production-ready Servo components into Gecko, resulting in the maintenance work being shared. So, basically I wouldn't expect any new player to enter the market in at least the next ten years.

steveklabnik 9y ago
In my understanding, the Servo team has never focused on ACID 3, instead prioritizing features that are actually used by the top X websites.
dman 9y ago
Thanks for a detailed response!

itomato 9y ago

Read up on the history of Konqueror and WebKit to get an idea of what it takes on the render engine side. Marketing, extensibility, feature-parity are in another realm, and remain a moving target.