Hacker Newsnew | past | comments | ask | show | jobs | submitlogin
Google's playing the closed-open game again, this time with HTML5 (extremetech.com)
107 points by mrsebastian on July 27, 2011 | hide | past | favorite | 62 comments


It is clear to me that this is a Google Chrome Experiment, from their labs, and that they did not had any intention of it working on any other browser but i got curious...

i was doing some research on it and they appear to be using sync xhr calls which causes FF to hiccup even after all this time.

https://bugzilla.mozilla.org/show_bug.cgi?id=383304 https://bugzilla.mozilla.org/show_bug.cgi?id=313646

they are marked as resolved but i just confirmed the xhr sync calls dont work in some cases.

google is using the sync calls here http://www.allisnotlo.st/static/js/API.js

Here you can try this to try and play the video on FF https://gist.github.com/1109890

you will notice your console throws some errors around line 42 of that file, if anyone knows what else is wrong here please let me know.


That would explain blocking of Firefox, but they're blocking Safari and WebKit nightly and... everything else that doesn't have "Chrome" in User-Agent string: http://yfrog.com/z/kh1rylsqj


i just tried accessing the app with Safari and it came to my attention that its trying to load mp4 and mp3 files instead of the webm videos http://yfrog.com/h7x8xup then i got even more curious and found this file http://www.allisnotlo.st/static/js/VideoList.js look around line 236

this.src = src.replace(/(\.webm+)$/, (navigator.userAgent.indexOf('Chrome/') == -1 && (navigator.userAgent.indexOf('MSIE') != -1 || navigator.userAgent.indexOf('Safari/') != -1)) ? '.mp4' : '.webm');

but looks like they didn't upload any mp4 files :( sad


None of this '90s-style sniffing is necessary: HTML5 supports multiple <source> elements allowing browser to choose and has canPlayType('video/webm') for JavaScript.


Huh? Where do you see a sync XHR in there?


there is none, my mistake


Er... there is no synchronous XHR in the snippet you pasted. All calls to open() pass true for the third parameter.


You are right, the error i get on my console isn't related to that, i misunderstood the error i get, i tried replicating today and the xhr request goes trough ok, http://yfrog.com/gysa24p makes me wonder what was my original problem, ill try to stop commenting by impulse.


this article is written with the obvious intention of bashing google, the writer didn't even do his research because honeycomb 3.2 source is available.


Where? http://source.android.com/ just says 2.3 is available, and google isn't turning up anything either.

EDIT: looks like the GPL/LGPL parts have been released (according to news reports, anyhow). Can you point me to a full buildable android 3.2 repo?


Only the GPL/LGPL parts are available, but the Honeycomb code isn't viewed as ready for public consumption yet. My guess is that we'll finally see the code when Ice Cream Sandwich rolls out.


Some openness. Waiting for corporations to deem it fit for public scrutiny but not public usage? Makes me suspicious if they released all the GPL'd parts or if they missed some.


Sure, but that was exactly what the author meant.


Oh, okay, so that makes it okay for the site to be Chrome-only despite Google's constant proselytizing over "openness."


"this article is written with the obvious intention of bashing google" i agree, it is way faster to develop for just 1 browser, plus we should be celebrating people are trying to break ground doing this awesome videos, i can only begin to imagine all the impressive work that what went into this video, good one google +1


I hope this was sarcastic, unfortunately I fear you're actually serious.

Scary.


out of curiosity why?


I tried to trick it with User Agent Switcher, but no go. It's not just examining the user agent -- it's looking for window.chrome, too.

Having watched the video and the word-spelling at the end, there's nothing about this sort of content that strikes me as stuff that can't be done in other browsers. And isn't that supposed to be the point of experiments like this?

Google should be showing us stuff that can't be done in any browser but Chrome because of those other browsers' technical limitations or deficiencies ... not just because they have the wrong user agent.


Google showed that this is Chrome showcase, a Chrome experiment, isn't that clear ?


Oh, they certainly have a right to call it a Chrome experiment, but the question is: What unique thing about Chrome are they trying to show off in this "experiment"?

Google could just as easily call ...

    <html><body><p>Hello world!</p></body></html>
... a Chrome experiment, but that's not going to make me switch from Firefox.


Yes, we know. That's the point. They shouldn't have because there was no reason to.


Disappointing article, given all the other ways in which Chrome is harming the web. Example: There's a bug on a site I made in Chrome. It is categorically incorrect behaviour on the part of Chrome, it breaks the specs in order to be faster. How do I fix the bug? By calling a proprietary interface in Javascript that only exists in Chrome. Yup, that's right - to make my standards-compliant web page work, I have to do browser-specific stuff for Chrome. It actually works fine in Internet Explorer 9, and has no IE9 (or 8 or 7) specific code in it. This is actually the second time that Chrome has been the stand-out browser for causing issues when building a site (unless you count Opera - I don't).


Have they acknowledged the bug and refused to fix it?


One of them is already getting attention so hopefully they will notice (you know how it is), the other one has literally only affected me as far as I can see and I haven't put together a minimal test case yet. It's a weird one!


Agreed, it would be nice to see the bug. AFAIK they take standards compliance pretty seriously. I'm sure someone would look at it, especially if Chrome is behaving differently from FF/IE9/Safari.


> AFAIK they take standards compliance pretty seriously.

As long as it doesn't interfere with their performance benchmarks.

http://weblogs.mozillazine.org/bz/archives/020267.html is an example; that issue is alive and well in WebKit, as are various other "optimizations" along similar lines.


Do you have a link on crbug.com?


  This is precisely the same kind of trickery Apple 
  pulled with its HTML5 Showcase for Safari, which 
  locked out other more compliant browsers in order 
  to make Safari shine line a beacon of hope for open 
  web standards. But Google isn’t building these 
  experiences to make the web look good: it’s building 
  them to make Chrome look good.
I don't get this bit. Isn't Google doing the same as Apple, then?


Remove the word "But" and it makes logical sense.


It's quite possible that there's a valid technical reason for this. Codec usage? Something bleeding edge to do with HTML5 Audio?

A slightly crappy but plausible reason is that they didn't want the extra hassle of cross-browser testing and pool of browsers capable of running this was pretty small.


Or the simplest explanation is that they wanted people to install Chrome to be able to see it.


The explanation is of course that they want people to run Chrome no matter the cost.

They can force people to upgrade to chrome that way. It's a well known way of doing stuff.

Microsoft has done that for ages in the past and got bashed for it until they started to fall.

When Google does it, it's ok tho. Figures.

ps: this is actually doing evil


Whatever the reason, the implementation is broken — it sniffs User-Agent and blocks everything else than Chrome.

If I change Chrome's User-Agent string to Safari, then this site blocks even Chrome:

http://yfrog.com/z/kh1rylsqj


It's a shame that an exaggerated article about a browser-sniffing art project that Google worked on is bound to get more attention from the public than, say, the contemporaneously front-paged [open sourcing of LevelDB](http://google-opensource.blogspot.com/2011/07/leveldb-fast-p...), which is a far more significant step in the direction of openness.


This is not an isolated incident, and goes to show the potential damage Chrome can do to the web. This is very similar to the behavior Microsoft exhibited with IE6, which left terrible damage for years after its release.


Useless article. This has nothing to do with "openness", just marketing: what is Google's incentive to building the site in the first place if they can't make people download Chrome to view it? That said, they could at least add a "Try it anyway" button for other browsers.


Chrome is free. Google builds websites to attract advertisers. Which makes Google's lack of openness all the more perplexing.

And what's this about not being able to get the Android source!? I'll buy an iphone right now if that's true. At least Apple doesn't try and pretend they're not evil.


Android source for Honey comb is indeed not available and hasnt ever been, so far. Only the GPL parts (kernel, iptables basically) are open because they're forced by law.

I don't think we'll ever get HoneyComb source. We might get Ice cream sandwich source. Might. Only Google knows.


I don't see how this makes Google's lack of openness perplexing at all. On the contrary, I would be surprised (and pleased) if they were at least as "open" (which I don't think is the right word anyway) as to include a "Try it anyway" link as they did in ro.me


They had a "Try it anyway" link for "3 Shades of Black": http://ro.me


You only think it's "useless" because it bashes Google, obviously a company you like. You ask what Google's incentive is to build the site if they can't make people download Chrome to use it, which is a bizarre question that seems to admit that Google is not about open standards at all. If Google was all about open standards, it wouldn't matter which browser you used.

The reality is that Google is not some benevolent open source company as they're often portrayed but are a gigantic advertising company with a financial incentive in getting you onto their products, which gives them data to index for advertisers. Chrome, for example, automatically interfaces with the Google search engine by default. They don't care about openness--that's a feel-good buzzword they use to make techies like them.

Google defenders have gotten worse than Apple defenders when it comes to justifying absolutely every shitty move Google makes.


Your comment is full of ridiculousness. Please stop.

I wasn't making the claim that "Google is about open standards at all", or that they are "some benevolent open source company". I would argue that they are not. I was in fact pointing out the author's own misunderstanding of this, as they seemed to expect Google to "be about open standards" and what not, which is ridiculous; they are a business, you should expect them to try to make money. None of what I have said is in any way "justifying" anything about Google, merely providing an obvious explanation for a trivial situation which the author of this article apparently couldn't grasp.


Ironically this article is presented in yet another degraded "iPad" interface I had to break out of to actually read the article.


I can't agree more with you. I hate these stupid iPad-optimized themes which in reality make the reading experience worse.

Why would anyone prefer faked scrolling or faked pagination (that's what this theme does) over the perfectly good native scrolling? I simply don't get it.

The cause seem to be these guys: http://onswipe.com/

They created this theme which seems to have infected half of the web's wordpress blogs. I wonder if any of those guys who install this theme on their blogs actually have an iPad.


All I get out of this is that Google calls their web browser "Chrome", while Apple calls theirs "HTML5".


I don't get why this is downvoted. You are right. At least Google explicitly marked this as a "Chrome experiment" instead of labeling it as a "HTML5 showcase".


How are you seeing the vote count? I don't see a way in my profile to turn it back on

[Edit to add] It's a serious question. Parent refers to downvotes of GP post. That makes me wonder if there's a way to turn the numbers back on. Something I would very much like to do.

Why the downvotes?


I'm guessing that at one point the post's text was gray due to downvotes.


Maybe because Google constantly markets itself on its use of open standards.


If that's all you get out of it, then you're intentionally ignoring the years of talk about "openness" from Google and its criticism of competitors who don't adopt open standards.


meh -- this is not an assault on openness, it's a silly little music video.

Locking other browsers out of gmail, gmaps, youtube, etc would be an assault on openness.


That would have been a much bigger assault on openness, to be sure. But this is quite bad as well - it clearly shows that Google cares more about pushing Chrome itself than pushing the open, standards-based web. As the article says, when Microsoft's web demos are more standards-based than Google's, you know something is messed up.

Not so long ago I don't think Google would have done such a thing. Apparently though there are changes happening.


Eh. They did a vanity site to promote a product.

When Goog starts adding /marquee/ tags and VBScript to their browser, I'll start getting worried about "openness."


"Goog" has been doing just that (except calling it "CSS Animations" and "NaCl", say), in case you missed it.


Chrome does not support <marquee />? Interesting.


Chrome supports <marquee>. cowboyhero was making an allusion to the browser wars[1].

[1] https://secure.wikimedia.org/wikipedia/en/wiki/Browser_wars#...


WebKit e.g. (Chrome & Safari) doesn't support the <blink> tag; though it can be recreated quite easily using CSS transitions. Not that you should though.


I don't get the fuss. It's a website that says "Chrome experiment" made by Google Japan in collaboration with some artists. It is clearly a promotional piece. The creators have every right to do whatever the hell they want with it. This has nothing to do with the openness of Google or Chrome. It is a demonstration of HTML5 concepts IN CHROME. Not in Firefox. Not in IE9. Not in Safari. This is why it's a CHROME experiment and not an HTML5 experiment.


Funny that EVERY BROWSER VENDOR out there is playing the same game.

Funny how, for web developers, "standards" is not about something being actually fucking standard.


I found this especially odd, since the Rome music video gave a similar warning, but let you continue and worked fine in Firefox 5.


So now we think that because Google wanted to lock the presentation of this video in a very specific browser so that it runs in the very specific way they want to without dealing with every minor complication that can occur in every one of the 98743 browsers that "support" HTML 5 and CSS 3, it means that they are sending a message about locking up the web for themselves?

This is purely sensationalist bullshit, reading into actions that have no ulterior motives.


This is clearly says Chrome experiment, so this is sensationalism for me. The HTML5 spec is still in editor's draft and browser vendors have different interpretations in some areas. Video is an issue but criticising Google when they open sourced the VP8 codec is unfair. If I were a journalist I'd focus on patents as blocking the open web and innovation rather than fabricate an agenda like this.




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: