You can download it, paste it in /Applications/Safari.app/Contents/Resources, and change the icon name in info.plist if you want...
Got a tip for us?
Let us know
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
Apple Looking to Boost Safari JavaScript Performance with New Accelerator Upgrades
- Thread starter MacRumors
- Start date
- Sort by reaction score
You can download it, paste it in /Applications/Safari.app/Contents/Resources, and change the icon name in info.plist if you want...
So when I launch the usual Safari icon from the dock I am using Safari 7, but if I launch the gold version it will be nightly? They share bookmarks, but are they still two separate programs?
Also, is nightly any less secure? I enjoy testing software, but I like to check these things .
Yes, the gold version is nightly. It's essentially everything that is Safari, but updated engine. The bookmarks, I think they're still there. I don't do bookmarks, though. It isn't less secure.
I have noticed that it doesn't do well with "start from last session". It doesn't really do it, I have to choose to "open from last session".
Also, it can be set up to remind you to update it. That's always nice. Or you can choose to update it yourself.
Is that sarcasm or serious? I have no idea. But FireFox lacks so much and updates so slow compared to Chrome. I appreciate google on that front.
----------
Do sync features like Reading List still work?
Thanks for sharing your wisdom! I'm going to give it a try. I can uninstall it like any other app, right?
In that case, I'd leave it to your thoughts on 100 tabs. I can't say anything about 100 tabs, I have never ever use more than 10 tabs at a time. So for me, less than 10 tabs, Safari is absolutely fast and responsive, no issues with its RAM management. I work in the same 5 tabs all day long and it has never eaten up that much memory either.
Since Safari 6/7, I haven't seen Safari eat up that much memory compared to other browsers. Safari 5 was a horrible release, the same 5 tabs would've eaten 10GBs easily.
The other thing is that I only use 2-3 extensions. If you have several, I'm guessing Safari sucks that much more compared to other browsers. I just know from experience that if I had the 5+ extensions in Firefox and Safari, they'd take up 200MB each tab but that was a while ago before I decided to kill some of the extensions.
6 Tabs opened in Safari 7.0.3: ~750MB
I replicated the same thing in Chrome: ~850MB. I'm not sure why there are more processes than Chrome though:
No problem.
And yes, you can. It should come up as WebKit. Also, make sure to put it in your applications folder if you'd like for it to remind you to update every so often. Otherwise, well, apparently it can't get the permissions it needs or some such thing.
How are you testing it? Compiling the source with the command line options or just using Webkit nightly?
It was just a quick test on sun spider website, I don't have time for more thorough testing right now.
No, what I mean is how are you getting the latest Safari that has FTL compiled in it?
Oh! No I don't yet. I tested the current nightly build against the current Safari with mavericks. However, I believe you can access the FTL via flags.
.I don't think this is entirely correct. Apple certainly has private APIs but was never obliged to provide Safari components to other apps. UIWebViews provide developers with easy web page rendering but aren't optimised and don't deliver the performance of Safari. Also using them would be a lazy option for a browser developer who should really incorporate their own web rendering technology rather than re-packaging Apple's. Google's Chrome uses it's own V8 JS engine, not Nitro so why they haven't built Chrome for iOS on this is probably more indicative of them providing a second-rate experience for non-Android users just as they tried to do with Google Maps.
Google Chrome in iOS does not use V8, it uses UIWebView and the reason is not lazyness, they simply have no choice. Apple in iOS for security reasons does not allow mapping RAM pages as executable, as it would allow execution of unsigned native code. This means not only Google Chrome, but no app in iOS altogether can implement a JIT compiler, Safari being the exception.
Until Apple provides the same JIT compilation capabilities in UIWebView too or allows third-party apps to implement their own JIT compiler there is no way for third-party apps to compete with the javascript execution performance of Safari.
This is a shame. I understand the technical reason on why this is the case, but knowing Apple I have to wonder if it's an intentional push for native over web apps.
This sounds interesting; I've never been a fan of the idea of asm.js, as it's not really proper Javascript at all, and developing in it is a pain. It's basically just a crutch for people looking to port code from other languages, and shouldn't really be used IMO, as it's more like assembly than a proper coding addition.
I'm hoping this is more along the lines of what Safari/Webkit is going to get, probably an extra optimisation pass that gives even better performance of more often used parts of a script. After all, really a Javascript developer shouldn't have to do anything different (within reason) for browsers to gain better performance.
That said, some kind of hinting mechanisms would still be nice; Chrome has a documentation standard that can also define the type of variables to be validated by the compiler (with an appropriate option set). But this kind of thing could be extended to help compilers to optimise variables and code segments sooner, or in particular ways (where supported), but wouldn't require any major changes to coding style like asm.js does, assuming you want to work with it directly of course.
I'm hoping this is more along the lines of what Safari/Webkit is going to get, probably an extra optimisation pass that gives even better performance of more often used parts of a script. After all, really a Javascript developer shouldn't have to do anything different (within reason) for browsers to gain better performance.
That said, some kind of hinting mechanisms would still be nice; Chrome has a documentation standard that can also define the type of variables to be validated by the compiler (with an appropriate option set). But this kind of thing could be extended to help compilers to optimise variables and code segments sooner, or in particular ways (where supported), but wouldn't require any major changes to coding style like asm.js does, assuming you want to work with it directly of course.
It probably can but if things hasn't improved since Mavericks the frame rate will suck compared to running it in Windows.
And still works perfectly on PCs today
Full screen, 1080p Flash, very low system load. Never a problem
How nice for you. That's not how it works on non-Windows systems, which is why it's going away.
What? HTML5 has not been forked by anyone. What Google forked is Webkit like Apple forked KHTML
----------
Ahem, Apple doesn't allow 3rd party Javascript engines, anyone developing a web browser must use webui and the Apple engine
Oh, so you are comparing the regular WebKit nightly against Safari 7.0.3, not FLT-enabled WebKit nightly against Safari. Got it.
I was hoping to see if anybody did compile WebKit nightly with the FLT flags and see if it is faster or not.
you are all pretty much talking over my head....but Safari won't load certain pages that Firefox will. Safari should be their flagship as it is a gateway to all else...but alas it is clunky.