Slashdot Mirror
Mozilla Releases First Build of Servo, Its Next-Generation Browser Engine (venturebeat.com)
An anonymous reader writes: As promised, Mozilla has released the first Nightly build of Servo, its new browser engine. This is the first tech demo of Servo, which Jack Moffitt, Servo project lead at Mozilla, described to us a few months ago as "a next-generation browser engine focused on performance and robustness." Packages for macOS and Linux are available to download from here: Servo Developer Preview Downloads. Mozilla promises that Windows and Android packages will be available "soon." And because this is Mozilla, you can check out all the code yourself over on GitHub.
So why is Servo so bad?
Browsers are hard. I remember the early days of Firefox, then "Firebird." It was terrible, crashy alpha software and completely unusable for years. And that was based on a "mature" language; they weren't developing the implementation language in parallel. The "browser" problem today is an order of magnitude more difficult because a browser is vastly more complex than it was 15+ years ago; browsers must precisely implement a much larger body of legacy and contemporary "standards" and do so with excellent performance on a much larger spectrum of devices.
Why isn't Rust letting them develop Servo faster and better?
Rust only reached 1.0 13 months ago; most of Servo development has been based on a rapidly moving target while trying to hit a rapidly moving target. Other than the fact that Rust isn't miraculous — and no one has ever claimed it is — the current state of Servo doesn't really tell us much about Rust.
Maw! Fire up the karma burner!
Firefox doesn't leak memory like a sieve, and hasn't since somewhere around version 7 before they managed to find a miracle cure for the shitty addons of that time. Now you really have to have bad luck to get it leaking memory like a sieve: shitty drivers, shitty addons/plugins, shitty third-party apps plugging into it, etc. Anyone claiming to the contrary has yet to prove it, though they like to bandy the claim about regardless.
browsers must precisely implement a much larger body of legacy and contemporary "standards" and do so with excellent performance on a much larger spectrum of devices.
If that were "all" they had to do the job would be much, much easier than it is. The real problem lies in the clause you need to add: "... and also gracefully handle the astonishing and manifold varieties of bizzarely broken, nonstandard, evil, and often just plain utter shit-content that web servers routinely vomit forth when queried."