Slashdot Mirror
Safari Browser May Soon Be Just As Fast As Chrome With WebP Integration (thenextweb.com)
An anonymous reader writes from a report via The Next Web: The Safari browser included in Apple's iOS 10 and macOS Sierra software is testing WebP, technology from Google that allows developers to create smaller, richer images that make the web faster. Basically, it's a way for webpages to load more quickly. The Next Web reports: "WebP was built into Chrome back at build 32 (2013!), so it's not unproven. It's also used by Facebook due to its image compression underpinnings, and is in use across many Google properties, including YouTube." Microsoft is one of the only major players to not use WebP, according to CNET. It's not included in Internet Explorer and the company has "no plans" to integrate it into Edge. Even though iOS 10 and macOS Sierra are in beta, it's promising that we will see WebP make its debut in Safari latest this year. "It's hard to imagine Apple turning away tried and true technology that's found in a more popular browser -- one that's favored by many over Safari due to its speed, where WebP plays a huge part," reports The Next Web. "Safari is currently the second most popular browser to Chrome." What's also interesting is how WebP isn't mentioned at all in the logs for Apple's Safari Technology Preview.
no it uses Middle Out which was inspired by ...
Some drink at the fountain of knowledge. Others just gargle.
(*waits for angry mob with torches and pitchforks*)
FLIF improves on webp by another 14%.
Hope it stabilizes soon.
http://flif.info/
WebP is a freaking image file format, its that freaking simple. I mean 500 words of bullshit about "technology from Google that allows developers to create smaller, richer images that make the web faster.",
Literally all they had to say was WebP is like JPG except it compresses more. Thats it, no need to say anything else.
These days - aside from (hypothetical and inevitably blocked) ads - it's not images that are the root problem. It's the half gigabyte of javascript.
I'm sorry Mr. Went To School For Web Design, but the moving pull-down menus and dynamic sliding content and whatnot is just not needed (except to justify your career). When I visit a website, all I really need are maybe four buttons: "BUY OUR SHIT", "DOWNLOAD UPDATES FOR OUR SHIT", "READ DOCUMENTATION ABOUT OUR SHIT", and "CONTACT US FOR ALL THE INFORMATION ABOUT OUR SHIT WE CAN'T BE BOTHERED TO PUT ON OUR WEB SITE".
Skip all the embedded activity tracking, metrics, demographics and dynamic content and we could go back to the golden days when web pages were under a megabyte on average without images.
"Oh no... he found the
It seems that it's not entirely unparsable.
A perfect fit for Facebook, no?
On Macs I use Safari first, Chrome second and Chrome only on Windows. Safari runs circles around Chrome, so I definately wouldn't like Safari becoming "as fast as Chrome" as it would indicate making it slower than it currently is.
For those interested, I recently setup a comparison between different image formats, including WebP, Daala, BPG and JPEG.
Opus: the Swiss army knife of audio codec
Patented up the wazoo. Might as well use JPEG 2000 then. Guess why JPEG 2000 was never popular?
WebP support is nice, but pretty trivial. It's WebM (video) support that everyone has been asking Apple to add, for years now. You can bet iPhone users would like smaller YouTube videos, WebRTC video conferencing, etc., but Apple is holding-on tight to H.264 AVC as the only available video compression format. Any coincidence that they are among the companies earning patent royalties from patent licensing the format?
Apple is the one big hold out, preventing adoption of better, open and free video formats on the web. Though WebP is somewhat related, it doesn't get us any closer to WebM and the open web, which Apple is single-handedly holding-up. Absolutely every other major tech company has thrown their support behind the Alliance for Open Media.
See:
- http://pipedot.org/story/2015-...
- http://pipedot.org/story/2016-...
Slashdot gets worse every day... Pipedot: News for nerds, without the corporate slant
Safari and Chrome may have been derived from WebKit at one point. But they’ve since diverged quite a lot. Google Docs in particular caused severe memory leaks in WebKit. Those were fixed in Chrome, but Apple has never imported those fixes, so Safari web content processes will eventially eat all your memory if you leave Google Docs open for a long time.
The HEVC patent pools (plural!) are probably what's blocking adoption of BPG, which is based on an HEVC I-frame. But which royalty-bearing patent sidelined JPEG 2000?
Any video codec can in theory be turned into a still image codec by wrapping a new container around a keyframe's bitstream. JPEG is similar to MPEG-1/2, WebP is based on VP8, and BPG is based on HEVC.
macOS also has a hosts file. I haven't tried it in Wine, but I'm assuming it'll work because I don't think Delphi programs to manipulate text exercise any obscure Win32 behaviors.
Where's a good place to try things on a Mac without owning one?
A hosts file is adequate if you have just a handful of addresses, but it really slows you down if you get more. It's read from disk and parsed every time you resolve anything. Linearly. You really want to run a local DNS server that stores the data in a proper data structure. A zone file can also block a whole domain while a hosts file is limited to a single hostname per entry.
The creatures outside looked from Alt-Right to Antifa; but already it was impossible to say which was which.
It uses wavelet compression. It's a patent and litigation minefield.
"Microsoft is one of the only major players to "
What does that mean ?
How about putting in the full context which was "Microsoft is one of the only major players to not use WebP".
Looks pretty obvious to me.
There ain't no such thing as proprietary standards only proprietary formats. Standards are by definition open.
It's read from disk and parsed every time you resolve anything. Linearly.
Hosts data, like all other data, is loaded from disk initially but read from RAM once cached (far faster)
KiloByte's point is that the OS's hosts lookup is O(n), while a purpose-built local DNS server that reads a hosts file can manage O(log(n)) using explicit cache data structures. The speedup when you resolve a site outside your top 50 can outweigh the slowdown for switching in and out of kernel mode. And it'd have the same security benefits as your hosts file, plus the ability to use wildcards.
It uses wavelet compression.
"Wavelet" is just a fancy name for quadrature mirror filters (QMF). The ATRAC codec in MiniDisc audio used QMF back in 1992, making it prior art for general QMF patents expiring before 2013. Or are there more specific patents for use of QMF in images?
It's a patent and litigation minefield.
Each mine has a seven-digit number attached to it. Which numbers apply to the use of wavelets in JPEG 2000?
It's read from disk and parsed every time you resolve anything. ...
Which means it is in RAM
while a hosts file is limited to a single hostname per entry.
Which it is actually not!
Cost free eBook I read (by iBook/Kobo/Amazon/ObookO/Gutenberg etc.): "The Green Odyssey" by Philip Jose Farmer.
It's read from disk and parsed every time you resolve anything. ...
Which means it is in RAM
Yeah, page cache, but this wasn't what I'm talking about. On a modern SSD the speedup from in-RAM caching isn't that massive anymore -- and unless you mounted noatime, there's a write for every operation anyway, both to the journal and inode. And then you still need to read the file and parse it to find that entry; in the most likely case, ie, a non-blocked hostname, you'll need to parse the entire file.
while a hosts file is limited to a single hostname per entry.
Which it is actually not!
You can have multiple entries per line, like this:
0.0.0.0 facebook.com plus.google.com twatter.com goatse.cx
but a single entry applies only to a hostname, not to the entire domain below it.
The creatures outside looked from Alt-Right to Antifa; but already it was impossible to say which was which.
and unless you mounted noatime, there's a write for every operation anyway, both to the journal and inode.
But that is asynchrony from the event reading the info.
Anyway: there is no reason that a DNS implementation is faster as it has the same limitations regarding disk access.
but a single entry applies only to a hostname, not to the entire domain below it. :D
That is interesting. Don't even know if I never knew that or simply forgot it
Cost free eBook I read (by iBook/Kobo/Amazon/ObookO/Gutenberg etc.): "The Green Odyssey" by Philip Jose Farmer.
Anyway: there is no reason that a DNS implementation is faster as it has the same limitations regarding disk access.
A DNS server reads zone files just once, at startup. It can do so because it has a means of getting notified of updates (rndc, zone notifies). The data is then stored in an efficient data structure that takes O(1) to find an entry (or O(k) where k is domain's length if we care about this factor, it's sharply bounded). Even without file access inefficiencies, the best a hosts file can do is O(n) (O(n*k)) -- and we can't ignore such inefficiencies, as reading and parsing a file takes ages.
(Some server implementations do O(log n) (binary tree) search rather than O(1) (hash [hopefully], trie [guaranteed]), but that's good enough to still beat a hosts file by orders of magnitude.)
The creatures outside looked from Alt-Right to Antifa; but already it was impossible to say which was which.
Agreed.
But if you would put in numbers for the n's and k's and what the O calculus omits, lots of c's for the calculations you would figure: it makes no sense to to set up an DNS server for a private person just to block some domains or hosts.
but that's good enough to still beat a hosts file by orders of magnitude.) /. already needs 10 seconds to load, what would I care about 1 second delay? When I have to spend X hours to set up my own DNS to block some hosts?
If that is just a factor of ten, and the time delay is below 1ms, who cares? Or even 1 second, who cares?
Cost free eBook I read (by iBook/Kobo/Amazon/ObookO/Gutenberg etc.): "The Green Odyssey" by Philip Jose Farmer.