Performance and standards look like dominating work on Microsoft's next version of Internet Explorer. As ever with Microsoft, though, it's likely to be the former that not just trips up the latter but that also continues to sour Microsoft's relationship with the rest of the industry. President of Microsoft's Windows and Windows …
Can't see the problem
So MS are saying they're going to improve performance, adopt further standards (some, but not all) and add the ability to hook into some local windows API's to up performance on machines with decent hardware.
Seems perfectly reasonable to me
evidence based - just not the evidence the reporter likes
I hate it when people talk about "folks" - it sounds so false and folksy. And it's Microsoft, who we know fine aren't remotely folksy.
The 3 Es
Clearly, M$ still hold tight to their 3E strategy (embrace, extend, extinguish).
Ie8 gets 13/100 on acid3 on my machine - at least I think it does, it doesn't render the test well enough to read the result.
If you let it download the security restricted activeX control that it asks for it scores 6/100 - then reloads itself 30 seconds later to give 13/100
Opera scored 99 on the same machine and chrome claims 100.
If I don't have to actually render any content I can make a seriously fast browser. and with the new improved "don't even bother starting" feature I can even make it use less memory.
Here we go again....... :(
"...Even inside of SVG there's SVG 1.2, SVG Tiny..."
Because understanding the desktop, general purpose spec over the mobile spec is just *SO* difficult, poor MS, life is soooo hard for them </sarcasm> *baulk*
If Webkit, Gecko and Opera can all support multiple standards with a fraction of the resource of MS, how come MS find it so "difficult"?! More like all those HTML5 and related open standards directly compete with SilverShite (which still doesn't bloody work properly, thecomedynetwork.ca being a case in point).
MS will be MS
I didn't really expect anything different. I'm not as annoyed as some purists will be- I don't have a problem with browsers leveraging different plugins- but it is disappointing that we won't see any sort of SVG implementation. That would radically alter what's possible within web design. Sigh.
Still, despite the disappointment I'm still taking it as a positive- speed and HTML rendering compliance are my two biggest problems with IE.
Rendering performance is all very well but what difference does it make if your web page renders in 50ms or 150ms? This time is dwarfed by the time it takes to actually retrieve the data from the internet in the first place.
As you say in the atricle, the rest is typical MS - they have a pathalogical aversion to standards. It must have really pained them to implement TCP!
One point that is made that I DO agree with though is about the huge number of "standards" that there are. But this problem isn't limited to web browsers. There are far too many protocols that do pretty much the same thing, AV codecs, etc etc. We would be far better off with fewer "standards". But that isn't going to happen so may as well forget it.
Typical MS control-freakery ..
"" IE general manager Dean Hachamovitch told The Reg earlier this year Microsoft understands graphics are important, but: "You can go down W3C site and count different standards out there," he said. ""
Wonks. If they'd programmed it PROPERLY, they could add that functionality through external modules on load (think extensions in Firefox) whenever a standard would gain dominance.
But no, let's invent another non-compatible browser they're really not integrating and packaging with the OS and as a preferential tool beloved most by idiots, morons, spammers, crackers and criminals ... proof enough MS, for all their standards-humping and open-love, are still as stupid as they have always been
Fail, because IE deserves to.
Faster than IE 8?
Great, except that, according to the graph, anything else is also much faster than IE 8, which seems to suck royally in that respect. Anyway, it's now time for Wintards everywhere to hope that IE 9 maintains such performance even after all the bloat has been added to finish the thing.
Nice to see poor little Microsoft catching up to the giants of industry though.
MS can't win with HTML 5
HTML5 is not a standard yet. If MS implement something as part of HTML 5 and it doesn't become a standard, they'll get slated heavily.
Bringing the rendering engine up to standards compliance is to be applauded and will make a lot of developers lives *slightly* easier.
The rendering engine isn't the only factor, especially with more complex or security oriented web apps.
The main thing they need to address is their security model. I can see the logic behind some of it, but it can be terribly hard to deal with. I've had cookies that were intended to make the site more secure blocked by the browser, and the countermeasures to deal with that make a mockery of the intended purpose of the cookie blocking anyway.
Conversely, we wouldn't have AJAX, the beloved "Web 2.0" technology without the MS Outlook team getting XmlHttpRequest built in to IE.
Hey ho. Dealing with this stuff keeps us employed!
Fair comment, except for COM & DCOM
Having developed with CORBA, COM & DCOM I can say you got that one very wrong.
While M$ was waiting for CORBA to provide that “Common Object Request Broker”, COM didn’t even have a name (it was just a protocol to enable OLE).. what changed things was the advent of “Component Based Development” and the need for something more generic than VBX. The irony being that CORBA was irrelevant before it was even finished.. because it is too slow for CBD (messages being routed through the ORB – COM used a proxy pattern & just got out of the way).
DCOM started as object-RPC over the Open Software foundations DCE stack.. ten years ago you could use DCOM to tie a Windows client to a Linux service (albeit without security), just about the same time that CORBA was “standardising” on a single wire-format “Internet Inter-Orb Protocol” IIOP..
CORBA’s not been killed by D/COM, it’s been killed by IBM and all the other J2EE vendors that have squashed it out of their app-servers to improve performance.
Microsoft’s extension to Java was consistent with other vendors extensions.. and better than coding JNI.. where they sinned big-time was excluding JDBC from VJ++.
Don’t get me wrong.. M$ attitude to HTML5 sucks.. but there are some cases where M$ "inovation" was welcomed by one and all (ODBC, far* pointers) & D/COM should be up there.
You were doing so well ...
... until your closing paragraphs.
"It's classic Microsoft. It's the attitude that led to COM and DCOM instead of using Corba"
WRONG: MS chose not to back CORBA because IBM was bullying its way around the standardization effort and making CORBA overly prescriptive in parts (that didn't need it) and woefully relaxed in others (that did). COM/DCOM won the distributed object wars. CORBA lost. Who was right?
"and that saw Microsoft "tune" Java and distribute a version not compatible with Sun Microsystems' implementation"
Sun won the case with a VERY weak ruling. Sun quickly released JNI so that they could start the complaints and push MS into court because they realized they'd given over too much power to MS, including the right to augment Java with platform-specific enhancements.
Frankly, MS' legal team probably did more damage to the case than MS did. A proficient team would have won the case hands down in MS' favor.
"Now it seems for all the talk of having learned its lessons on standards in IE 8, IE 9 will be a case of moving slowly and selectively where it suits the company and using its own approaches elsewhere."
WRONG - they're weighing up time vs. resources vs. features. There are literally hundreds of "standards" currently either boiling away hoping to become "real" standards in the future Imagine you run the IE team. You have $100 - where do you spend it:
b) Improve rendering performance
c) Improve compositon support & better compliance with CSS
d) Improve security & reliability process isolation & extension sandboxing
e) Improve the remaining top 3 customer pain points based on actual telemetry/stats and customer feedback
f) Implement support for any number of random other technologies which may never gain widespread use
I'm guessing youd spend VERY little on f.
What would be the point in building complete support for a version of HTML which hasn't yet been ratified? How many sites are likely to widely use the new HTML flavor before SP1 can be rolled out?
"It's an approach that will continue to leave developers struggling to support multiple browser architectures and invariably defaulting to Microsoft first and everyone else second based on market share, while leaving standards advocates and browser rivals as frustrated and angry as ever."
No two browsers support every standard in the same way. There will ALWAYS be differences that web developers have to code around. Until the W3C does a better job of accurately defining standards and producing *real* compliance test suites this problem will persist ... even if MS closed the IE team down tomorrow.
Courtesy of Wiki's page on the test:
IE9: 32/100 (so far)
So what could they aim for?
Firefox 126.96.36.199: 56/100
Konqueror 4.0.2: 61/100
Konqueror 4.3.2: 89/100
Android 1.5: 93/100
Firefox 3.5.5: 93/100
Firefox 3.6: 96/100 (so far)
Chrome 188.8.131.52: 100/100 (not pixel perfect rendering)
Chrome 184.108.40.206: 100/100 (not pixel perfect rendering)
Epiphany 2.28: 100/100 (+ pixel perfect rendering)
Midori 0.2.0: 100/100 (+ pixel perfect rendering)
When expectations are low, improvement is easy
"IE 9's score already beats IE 8 on Acid3"
Well that's hardly difficult. IE8 on Win7 here barely struggles to 20/100.
Come on, Microsoft, just give us SVG. Please? Pretty please? We've been wanting to use that since IE 4, for Christ's sake.
HTML is NOT a Standard
HTML 5 is not a standard so it's not unreasonable that Microsoft hasn't implemented it and, furthermore, they have comitted a bunch of people to help work on it. Meanwhile, they're trying to improve the standards compliance, performance and visual appearance of their own web browser.
You have to be pretty odd ( or working for one of the other browser manufacturers ) to have a problem with that.
Wow, that was quite a rant!
The only problem is that it seems to be predicated on the idea that the rendering improvements somehow involve MS-only 'extensions'.
Maybe I'm dumb, but from what I saw they are hooking into DirectX *within* the renderer - no HTML/CSS changes required, no new 'APIs' to code against....
@You were doing so well
You spend the $100 buying dinner for Fox and persuading them to make their site only work on windows.
Web browser war, reloaded
Yahoo's Douglas Crockford thinks the web needs a second round of 'browser wars' to push things forward.
Perhaps he'll get his wish.
To all who are claiming M$ shouldn't adopt HTML5 because it hasn't been ratified as a standard should remember that EVERYONE ELSE has agreed that it will be the standard (except for crApple who don't like the Theora video format and want to use their own... sound similar?)
I was writing web sites in HTML 4 before it was ratified and I'm really glad I did.
IE is slowly losing market share. It'll take a while but eventually all the dumb-asses out there will come across a good Samaritan who'll point them in the direction of ANYTHING ELSE!
Beer. All WebDevs will need lots of it soon.
slightly off topic but...
I loved the phrase in the article that said "There are about 80 plus standards out there."
Reminds me of some old joke about getting rid of the old stuff and creating a "new standard".
Sadly I can't remember the joke now.
Anyway "80 standards" is probably an oxymoron [I've always wanted to use that word]
Hey guys, it's just a browser thingy
You're free to choose from quite a few of them
Some work better than others
The Internet is like that.
Still trying to get it right? Get outta browsers and let the market leaders slug it out. That's where the innovation is coming from
"WRONG - they're weighing up time vs. resources vs. features"
How can Microsoft have the *most* amount of developers and cash and make the *least* standards compliant browser? Why did it take 5 years between releases of IE6 and IE7?
Stop apologising for them. They have absolutely no excuse for doing such a shoddy job.
Do Microsoft presidents speak in English?
" "We're getting very close to basically being a wash because by the time you get down to this amount of performance, the other subsystems are really going to dominate in terms of where the best places to improve are," Sinofsky told PDC."
I presume that "being a wash" (whatever that means) is not the same as "being awash", which seems more likely. As for the convolutions which follow, I despair...
Haven't Microsoft heard of the command pattern?
Most of these new standard use XML name-spaces, so are easily identified.
Use the command pattern and the extra rendering standards can just be add-ons to the rendering engine; that way you can add each standard, and version, as new rendering extensions, simples!
If it don't work in Firefox, under Linux, I just go somewhere else.
Someone kick Adobe for an updated 64 bit Linux flash please...
I call CHEAT
...on the Sunspider data bar chart. I don't have time to read all the comments so far, so apologies if this has already been posted.
The Y axis has been extended by inclusion of the dismal IE7 to make it look like there are minimal differences between other browsers (including IE9)
An honest presentation would use FF 3.5 to define the Y axis. We could then eyeball both the absolute and percentage differences between browsers. As it is, we can just squint and guess, or just accept the gospel according to Micro$oft.
I'm still not going to use it.
Re: Pragmatism before standards
"Only purists would complain MS isn't doing enough. Ask any non-geek what improvements they would like to see in IE and they'll say "make it faster"."
That's because the non-geek doesn't realise that the bottleneck (by at least an order of magnitude) when they surf the web is pulling the bits down their ADSL line (whether it is capped or not). Maybe if some of the graphics were fragments of SVG, inlined in the body text rather than requiring a separate HTTP request for a (larger) JPEG or PNG file, *that* would do more for performance.
But then, only a purist would care about that, and "everyone knows" that the best strategic decisions in any field of engineering are made by people who know jack shit about the subject.
Do we really NEED yet another version of this thing ?
I'm all for continuous improvement, but surely by IE8 they had the thing sorted.
Oh wait ....
Most business apps will only run on IE6.
So client only run IE6.
So I have to deal with IE6.
As for not fully supporting HTML5 and "special" doo-hikey APIs, the pain will just continue.
"not invented by Microsoft yet part of the W3C's HTML 5: Scalable Vector Graphics (SVG)"
Correct me if I'm wrong, but when did SVG become part of HTML5?
SVG = XML image format
HTML != XML
They killed the idea of images and the markup being integrated when they stupidly decided to kill XHTML
On another note, SVG competes with Silverlight too much for them to implement it. Bastards.
"being a wash"
While I don't know the origin of the phrase, it means that the end is no different than the beginning, that there is no net gain.
re: AC @ ~12:51 yesterday
They didnt' implement TCP/IP properly either. Or they implemented parts of it in IE8 because when I update my local DNS, I have to 1) ipconfig /flushdns and 2) restart IE8, otherwise I have problems connecting. If I could run Linux at work, I'd never touch windows again.
"They killed the idea of images and the markup being integrated when they stupidly decided to kill XHTML"
XHTML isn't dead. If you read the HTML5 standard, you'll note that there are two versions; HTML5 and what they're calling XHTML5, an XML rendering of HTML5. And that's where the SVG and MathML support stuff are expected to be used.
A voice of reason amongst so much noise.
I think you give Microsoft too much credit with regards to their resources...
They probably outsource a lot to India, Pakistan and other second world countries to save on the immense costs that these systems require for design, development, testing, etc.
As such, how do you expect high quality when the teams are so far apart, everyone talking a different language, standards are different, etc.
Reminds me the tower of Babel...
- The land of Milk and Sammy: Free music app touted by Samsung
- 20 Freescale staff on vanished Malaysia Airlines flight MH370
- The long war on 'DRAM price fixing' is over: Claim YOUR spoils now (It's worth a few beers)
- Dell thuds down low-cost lap workstation for
cheapfrugal creatives or engineers
- NSFW vid LOHAN chap hooks up with busty stratominx in cosmic pleasure cruise