6 ms·
Every time I hear about Servo, it mostly can be boiled down to: A: We did a ton of work B: We're not even remotely close to something really useful yet I thi
by vanderZwan 12y ago
Every time I hear about Servo, it mostly can be boiled down to:
A: We did a ton of work
B: We're not even remotely close to something really useful yet
I think that says more about how websites work, to be honest.
- jerf 12y agoThe stack of web technologies is staggeringly enormous. It may not feel like it just scanning over a simple HTML page, but it is, and there's a lot of invisible-yet-critical stuff like proper same-origin policy enforcement. The miracle is not that it takes Servo a while to be created, it's that this stack works at all, and that they're even trying!
- higherpurpose 12y agoCouldn't they just do something similar to Microsoft's Edge/IE11 approach? Servo for HTML5+/CSS3+/Javascript ES6+ and then keep Gecko as well inside Firefox for "legacy websites"?
- adapt_throw 12y agoIf the point is to make a more secure browser, allowing a webpage to opt-in to use the old, less secure engine instead is probably a non-starter. (In any case, I think the complexity is fundamental to many of the specs involved.)
- TazeTSchnitzel 12y agoHTML5 is no smaller than its predecessors. It's considerably larger.
- gsnedders 12y agoHTML5/CSS3/ES6 define how to process legacy content as well as current — the only actual branching for versions is quirks mode and that has relatively minor effects which aren't that much code to implement. Edge should, in principle, support everything, including legacy sites (at least as well as other browsers do today). Trident is only being kept around for sites that do extensive UA sniffing and rely on obscure behaviour — and pretty much only for the intranet.
- Manishearth 12y agoGecko is not exactly separable from Firefox (unlike Blokc/Webkit and I'm guessing Trident/Spartan). The firefox UI depends heavily on things like XUL and XPCOM and whatnot which are nonstandard features in Gecko and would be a pain to support in Servo. I guess you could run two browser engines -- one for the UI, one for the content, but I'm not sure how well that would work.
- plorkyeran 12y agoFireFox using some libraries written for Servo is a nice deviation from that. The sheer amount of work involved in getting Servo to a usable state should be less surprising when you consider that all of the browser engines in current use date back to the 90s in some form (1997 for Gecko and Trident, 1998 for KHTML/WebKit/Blink).
- sp332 12y agoNot only that, but Servo is experimenting with new techniques for just about everything as they go. Progress may seem slow if you measure how "done" it is, but it's much faster than other projects if you measure how much research is being produced.
- yoodlepipping 12y agoResearch? I don't think they've published a single paper have they?
- ffk 12y agoBreakthroughs don't always result in research papers. They sometimes end up in products instead. :)
- pcwalton 12y agoThe parallel algorithms that Servo uses to implement advanced features of CSS are, as far as I know, novel (block formatting context size speculation, float impaction detection, absolute positioning traversals, parallel automatic table layout, parallel border-collapse).
- vanderZwan 12y agoI vaguely recall them discovering a number of bugs in the spec along the way as well.
- pcwalton 12y agoYes, we've discovered several spec bugs, as well as bugs in other browser engines.
- Bjartr 12y agoPractical research, not academic research.
- 12y ago
- reddotX 12y agothe B is wrong :> they are getting closer and closer to gecko rendering level. gecko vs servo (rendering a couple of sites) https://www.youtube.com/watch?v=pZGhnqtXVdc https://www.youtube.com/watch?v=pZGhnqtXVdc