These Weeks in Servo 34 | Hacker News Reader

These Weeks in Servo 34

blog.servo.org

63 pointsManishearth24 comments

How usable is Servo currently? Is the web security model fully implemented?

Has anyone in the team switched to using it for all their browsing? If not, when do you plan to start?

Anyway, thanks for working on this! Servo is one of the most important software projects, since it will hopefully give us a secure or at least securable browser, and maybe give a chance for Mozilla, the most trustable browser maker, to get the #1 market share spot.

We're not, but we hope to be dogfooding by the end of the year.

Lots of Web features are implemented, but most of them have various minor and not-so-minor bugs. It's very easy to get involved in as far as Web engines go, thanks to Cargo, so feel free to build it and file bugs (or even help fix!) issues you see on your favorite sites!

Tasting the chefs cooking it much better than dogfood

> How usable is Servo currently?

... Ish. Works good enough for browsing (i.e. googling stuff and reading), but sites with complex interaction (eg Gmail) don't work too well.

No, security is not fully implemented.

> Has anyone in the team switched to using it for all their browsing?

Last semester I used it for all my web-based studying.

I'm also getting us close to be able to use it for Github things. Reddit shouldn't be hard either.

> Servo is one of the most important software projects, since it will hopefully give us a secure or at least securable browser

I'm very interested in Rust, and in Servo as a "killer app" that demonstrates its promise. But what if they go to all this trouble, and it's significantly slower than C++-based browsers? (Yes, Rust can be as fast or faster in theory, but C and C++ have so much more resources behind them.) How much is stability and security worth to you?

I see (admittedly abstract) tradeoffs made in favor of performance all the time in software where data integrity is supposedly paramount. Will we really make different choices in consumer software?

In their benchmarks it performs favorably when compared to Firefox's gecko engine: http://take.ms/LeUjm

  > C and C++ have so much more resources behind them

Since we use LLVM, we get all of the gains involved when Apple and other companies make it better. So that helps.

> But what if they go to all this trouble, and it's significantly slower than C++-based browsers?

Do you have any specific reasons why this would be the case?

I'm curious, why do so many pages (like this one) now have huge fonts and lots of empty space? That text is like... 3 times too big.

Makes it easier to read.

It looks really good on mobile.

So the Servo window by default only takes up a quarter of the screen, which makes some sites behave more like mobile sites. I picked this theme because it looked nice on Servo, so it's looking nice on mobile as a consequence :)

I actually want to update our CSS to something better (PRs welcome!)

Initially I picked this theme because it looked really nice in Servo, but we support a lot more CSS now and one of the nicer themes should work well too. I know that the older This Week in Rust theme works in Servo (they recently switched to Bootstrap which mostly works, but has some quirks), and I've wanted to switch to that for a while. Never really sat down to do it, perhaps I should do that next.

Just set the font size, line height and margins to sane levels and it'll be fine

Here I set font-size to 1em, line-height to 1.2em and margins to 1em:

https://i.imgur.com/BXDrg2a.png => https://i.imgur.com/hqHFNaf.png

On behalf of the few over-50's on this site, we find the first example much easier to read.

I'm just about to hit thirty, but the first example is still _much_ nicer to read for me.

That's a bit too far. Just changing paragraph font-size to 1.2em would be enough: https://monosnap.com/file/2IbdkolMW8nVIF1cGsX75BeiwR5b7k.png

Please don't make the font smaller. It's nice to be able to read text without squinting.

You have to be kidding me? Their paragraph font size is literally the same as a level 2 heading by default. It hurts my head just looking at it.

> 3 times too big

I know you exaggerate, but I just tried zooming to 33%, and it's really hard to read, even on my 15 inch retina screen.

75% was nice, though.

Cheers on using Reviewable for a better review process, but I notice it adds loads of useless text to the git commit messages[0]. I always figure that GitHub, Reviewable, and other hipster things might not be around one day, so I have stopped using GitHubisms like "Fixes #1" and such in my commit messages, and instead referencing the commits from pull requests and issues. That way, when GitHub is gone at least the git commit history still makes sense ("Grandpa, what is GitHub?").

[0] https://github.com/servo/servo/commit/4dc986bca35f986e2f15f1...

When Github will be gone (if that ever happens during your lifetime), will that extra clutter in your commit messages really matter that much?

Look at commit messages in the Linux kernel, that's what we should be aiming for—whether GitHub or GitHubNext will be around or not isn't really the issue, just an easy thought experiment to illustrate the point.

https://git.kernel.org/cgit/linux/kernel/git/torvalds/linux....

The merge commit of pull requests in Github contains the PR text. Tomato, tomahto.

Reply on news.ycombinator.com