I solved the embedded Chrome OS ad equation and won a Cr-48 | Hacker News Reader

I solved the embedded Chrome OS ad equation and won a Cr-48

sylvainzimmer.com

324 pointstonyskn88 comments

Now its time to figure out how to set up a server that can handle few thousand viewers.

Cache: http://webcache.googleusercontent.com/search?q=cache:http://...

I still really don't understand what's so hard about a couple thousand views. I've had a couple posts hit the front page (even #1 on HN and #5 on prog.reddit), and my server was fine. I use Apache to serve the resources and Clojure to handle web requests. Nothing fancy.

Wordpress without caching on a VPS that is probably not configured well. w3tc or wp-super-cache would probably fix his issues. Installing it while the site is hammered.... will be fun. Perhaps as challenging as solving the easter egg.

Apache redirect to a static page, while he gets the cache plugin set up?

exactly what I did... tried to switch apache to port 81 to be able to configure the cache plugin. But even this way, all the disk accesses were taking 30 seconds each...

The real error was posting it there in the first place and not directly on the Jamendo blog... Never again! :)

Any out-of-the-box Apache setup on, e.g., Ubuntu will, with absolute certainty, appear to crash if you subject it to several thousand pageviews in a short amount of time. KeepAlive is on by default, and that essentially amounts to turning any group of five or more interested people who share a 3 second window into denial of service attack.

This used to drive me freaking bonkers -- I was installing caching plugins, offloading static resources to other domains, tweaking the number of worker processes, etc etc. Nothing worked, and my blog regularly crashed almost every time it hit an audience larger than HN's. Turned off KeepAlive, bam, no problems since.

How hard would it be for them to make it degrade by default? ie. Use KeepAlive until it starts having heavy load, and then disable it when the load becomes a problem.

ab with 500 concurrent, no keepalives: Requests per second: 13428.16 [#/sec] (mean)

ab with 500 concurrent, keepalives: Requests per second: 14403.98 [#/sec] (mean)

Keepalives allow the browser to maintain a socket between the browser and server and request multiple objects. So, rather than having to do a connect/teardown for each object, you remove that overhead. Some versions of Apache in conjunction with php and mysql_pconnect had issues in the past that might explain the OP's issues with keepalives, but, it has piqued my curiosity. In general, keepalives should reduce communication handling and allow more throughput.

If keepalive fixed it, your configuration had other problems especially if you were using mpm-worker. email me at hnusername@hnusername.com - I'd be interested in talking further. Ubuntu and Debian distribute configs meant for home users/developers - those configs are certainly not made for production webserving.

You mis-understand keep-alive. Keep-alive is good and should be used. The real problem is elsewhere in your configuration.

Perhaps I should write a blog post about this.

Varnish, ahead of time?

From the screenshot of the secret page: "Also, we can only give you a Chrome notebook if you live in the United States and ... "

I find it such a shame that these types of conditions are so popular for most contests / easter eggs, etc.

Are the legal issue regarding this that complex?

I'm assuming Google is not really that cheap regarding postage costs, since we're also only talking about 1 notebook here.

Also interested how it all worked out for Jamendo, since (from what I see) they're based in Luxembourg.

It appears to be an issue with wireless regulations.

I doubt it. It applies to most giveaways of any kind (free posters, free calendars, etc). They usually only accept entrants from one country or equivalent legal entity. I expect it's because of variation in contest-related laws between countries, tax laws, etc.

Yes, in some countries if you advertise a public contest you first have to register it and vet it with your government, who then make sure everything is done fairly. That is something google simply cannot afford to do.

I am sure they will find a way to reward jamendo somehow. They'll invite the guys for a formal visit and give them business cards in the shape of netbooks, at the very least.

hi guys. I'm Sylvain who solved the thing, thanks for relaying the info here! :)

The blog indeed crashed, it's on a very small VM with a remote filesystem and even with a WP cache plugin.. not so much luck, even with apache stopped.

However our company blog has a proper sysadmin ;-) So after I setup a redirect there no more issues... until we get on slashdot maybe ? :)

Cheers!

From the screenshot of the form you got after solving the equation, it looks like you need to have a U.S. address to receive the Cr-48, and it would appears poor you're located in Paris/Luxembourg. You need someone to forward it on to you? ;)

yes that's also a funny side of the story. As I was going trough the border on tethering, with barely any cellphone reception, I was trying to fill out the "one-time" form without any U.S. address in mind ;-) Luckily I have a cousin in Houston so gave him a quick call to get his address and I submitted the form as soon as I got my data link back up :)

generally speaking, it's extremely frustrating for an european startup to see these regular U.S. only launches from Google. (come to mind Google Voice, ebookstore, google music, and now Cr-48...) But well, small price to pay to enjoy living in Paris most of the time :)

As a European working at Google (and there are plenty of us), I feel your pain. However, although so many things are easy to scale internationally, products with licensing and regulations are not in that category. We do work actively on making this less and less the case, though.

yes, indeed. I can imagine carrier or licensing deals being complex... that's why I said we were just frustrated, not angered against Google :)

Cool that you work on the Chrome team. Can you share more insider info with us about this story and its makeof ? ;-) thanks!

The equation - http://img545.imageshack.us/i/chromeequation.jpg/

Thanks for the link but the first few times I tried to load this page the image was broken. imgur is awesome, give it a try sometime!

Some symbols are hard to see, but I hope it is right.

LaTeX'ed version: http://yfrog.com/gyegnp

Yes, this is what I saw. HOWEVER, when you run he numbers, you can NOT get the same values that Google represents. The value for "C" is WRONG! It should be 8338041. This changes everything. ===> http://goo.gl/fpJsJ

From the article:

> most people I know including myself couldn’t live without Chromium/V8 anymore.

Is this a common sentiment ?

The statement is definitely a bit sensationalist. That said, yes. Even non-techy people I've seen checking out chrome comment on the speed and how much they like the UI.

I'll spend some more time with Chrome then, I have it installed but normally only use it for compatibility tests.

thanks!

It takes a bit to get used to if you're a Firefox 'poweruser' and I'm not sure 'Web Inspector' is quite to the maturity of Firebug yet... Still, it's my primary browser.

One cool thing not to miss is in the unified location bar, it does some search stuff via the 'manage' button in the basics pref pane. It auto-senses search fields on various websites and lists them there, then you can go in and setup a trigger for it (eg, wikipedia => W, google images => gi, etc).

Sir you have just made my day.

I'm not sure how common it is, but I share the sentiment. I find it incredibly painful to use any browser besides Chrome because of how slowly they run (even with all extensions turned off).

I can relate, though maybe he is being a tad sensationalist.

I expect (say) my bookmarks to be synced between all my computers. I'm sure plugins do this on other browsers, though.

The biggest reason I don't consider Opera or FF for daily use anymore: Non-chrome browsers drive me crazy due to the way they close tabs. It is very stressful compared to mouse-closing tabs in Chrome and its "smart positioning."

Using Chrome is like driving a luxury car for the first time. I remember when I got in a '93 Mercedes SL 500 in the rain and when I slowed down so did the windshield wipers. I know others replicated the behavior in later models, but driving that car made me realize that every other car was designed by people who didn't, apparently, drive cars.

I love using the 'smart tab positioning' example: how many times have people moved the mouse to close 4 or 5 tabs? Sure, it's not an impossible headache. But most of us do it regularly enough that it's nice to have the button in the same place for each tab. It's almost as if the designers used a web browser before. And it's great.

And hard to imagine giving up.

I love using the 'smart tab positioning' example: how many times have people moved the mouse to close 4 or 5 tabs?

I would expect that most people close multiple tabs with [cmd|ctrl]-w, precisely because of the back and forth mousing otherwise required. One thing I've worried about while using Chrome has been accidentally closing more than one tab at once, since, being a klutz, I don't have wonderful control over my index finger, and triple-clicking to select a paragraph is so common that I do it accidentally sometimes when intending to single click on a button or something. So far, I don't think it's actually happened very often, though.

If you accidentally close too many tabs, Shift-Ctrl-T will reopen them. And will remember content in forms.

I like Chrome.

Shift-cmd-t makes you wonder what the hell everybody else is doing that it took so long to get this feature.

Actually, its been in Firefox for at least as long as Chrome has existed... I can't remember a Firefox build without it, but... it's been a while.

You are, of course, correct. I don't know what I was thinking. I haven't really used ff since switching to chrome, although I find I do need to boot it up to make the occasional flash application work. Also, webkit-based browsers have some wonky css handling. I guess my brain has some sort of cognitive bias where I unknowingly attribute everything that's awesome about Chrome to Chrome itself.

Also, Ctrl-Z in Opera also works, which makes perfect sense and is less for the user to learn, and is discoverable (in the sense, they can go to the standard Edit -> Undo menu which is more visible.) Although actually, Opera also has a visible basket of closed tabs on the right of the tab bar, which is even better.

holy shit. you just made my day.

  I would expect that most people close multiple tabs with [cmd|ctrl]-w

If you think most people use keyboard shortcuts, we're working with completely different customers ;) Finding someone that uses the keyboard to copy/paste is decently common, but I rarely see use of shortcuts beyond that.

I find I use mouse close most often when I'm "info dumping", as mouse scroll and cmd+click to open a new tab are, for me, mouse operations. Rapid open/close of several pages is normal. Once I have 3-5 on a topic, I go "info diving" and nothing gets closed 'till the end, at which point cmd+w is how I tend to close out of the extra pages I had open. But cmd+w by even power users is, in my experience rare.

It would be great if the browser taught people these things during use. Kind of like some small drop down at the top "I see your closing a lot of tabs, have you tried ctrl-w?".

As it stands unless I'm wrong a user would have to go digging for these shortcuts online or by clicking help and actually knowing what they are looking for. Seems like a stretch for most browser users.

For a common task such as closing tabs mouse gestures should be used. There is no need to hunt buttons you just use your muscle memory.

How is mouse-closing tabs in Chrome different and what is "smart positioning"? Didn't notice myself.

When you have many tabs open in Chrome, the length of each tab shrinks so that they all fit in the bar. Smart positioning refers to, when you click a close button on one of the tabs, the rest don't expand to fill the remaining space until you move your mose away, so that the next tab's close button falls under your cursor in case you want to close it as well.

When you close a tab, tabs could become larger because there is now more space for each one.

But in Chrome they aren't resized until your mouse leaves that area. This way you can close multiple tabs with clicking, no movement required.

This design quirk is so useful and amazing how others haven't though of it before (once you see it it seems such an obvious idea) that some days I think about looking through the Chromium source and finding the name of the developer who created this interface feature so he/she could be properly thanked.

It was designed and implemented by Ben Goodger, the front end lead.

Compared to closing several tabs in Chrome, doing a similar thing in the Win7 shell (close several open folders) is infuriatingly difficult. The target size of the close icon is so small, and missing it is doubly bad: it makes you start the task all over again, plus a random window is now positioned to the front destroying your work context.

Eh? Window close buttons are larger than tab close buttons.

In Windows 7, you can hover over the taskbar icon for a browser and it will show a preview of all your tabs. Clicking on the tiny close button on a tab's preview will close the tab, but if you miss it, you will switch to that tab (and window) instead.

Thanks, this is the UI I'm referring to.

I also sort of wish I could press Alt+Tab, and have clickable close buttons in that UI.

you can middle click the text and it'll close it.

just like how you can middle close to close a tab in most browsers

I've been leaning back towards firefox recently due to how chrome handles large number of tabs: poorly. It'll squeeze them down to the point that they wont even show the favicon, making finding an open page annoying - compared to firefox, which has a minimum tab length, and scrolls the tab bar if it grows too large to fit.

The rest might be a personal prefference, but I rarely use the mouse to close tabs anyway, ctrl+w is faster, and I can hit that as many times as I want without moving the cursor anywhere. I also preffer the / and ' shortcuts firefox has.

Just curious, but if you have that many tabs open why not split them into different windows?

If it's split into windows, not only do I have to hover over every tab to find the one that I want, but also repeat the process for every open window.

The alternative is sorting the tabs somehow, but that's more maintance than a middle click.

That's why I use FireFox's Tree Style Tabs add-on.

Indeed - the lack of a multiline tab bar for Chrome is mystifying. No extensions do the trick, either.

true, it was a bit over the top. I've edited to "can't go back to any other browser than Chromium/V8 anymore." which is closer to my actual feelings :)

And this comes from a former hardcore Mozilla fan :)

I've barely used Chrome before, but shouldn't you use mouse gestures to close tabs in Opera? It's by far the easiest way, along with ctrl+w. No need to aim and click on a specific button.

That's a terrible reason not to use FF -- all you have to do is middle click anywhere on a tab.

I have been using Chrome since it came out but this month I am trying FF 4.0 and it's pretty good.

Could I live without it? Sure. In that way it's probably a little over the top.

As many replies to your comment suggest, the user experience is pretty great. However, in my opinion, OP was probably referring to the awesome performance of V8 (the Javascript "engine" under the hood). Chrome has become my defacto recommendation to anyone who uses Javascript-heavy web apps. There's also a nice set of developer utilities for Chrome that make it a very useful tool when developing web applications. Perhaps he's suggesting that, as a tool, he finds it indispensable?

Chromium/V8 specifically, no. Recent versions of Chrome/Chromium, Safari, or Firefox are all acceptable IMHO.

I would hate to have to go back to browsers older than 1 or 2 years though. Or IE.

24 occurrences of ";-)" — true happiness!

I dont get that : "we realized “900.91″ did actually reference the goo.gl url shortener"

I do not see how he did that?

You don't see any similarities between "900.91" and "goo.gl" ?

Just tilt your head a bit :)

    9 -> g
    0 -> o
    1 -> l

oh thank you! haven't realise that!

There isn't one standard way to convert a latitude / longitude coordinate into a x/y coordinate for a map. Google Maps was the first to use the Spherical Mercator projection. The EPSG (?) number used to reference the Spherical Mercator was originally 900913. EPSG:3785 now refers to the Spherical Mercator too.

http://docs.openlayers.org/library/spherical_mercator.html

Kudos. Too bad that special page didn't also include a job interview offer.

Well, now please make Jamendo a joy to use...

true... we're not happy with the current state of the jamendo.com website. 2010 was mostly focused on getting close to breakeven thanks to jamendo.pro (which we did). We also did a management buyout from our former VCs, laid of a lot of staff, so all in all, we're just happy to still be there, growing, and hiring again :)

Now we have several ongoing usability projects, we released a HTML5 player a few weeks ago (replacing the old Flash crap) and are now working on layout and stability improvements on the main pages. It will definitely make Jamendo a complete joy to use again, if we can just spare some time from breaking Google easter eggs ;-)

cheers!

I think the site is great! One minor annoyance: the volume gets reset to about 75% when I close the player completely and re-open it (but not when I switch to a new album with the player still open). The problem with that is that it is too loud for my setup, and I end up having to lower it each time. It would be great if it could remember its last value.

thanks p206.. we'll fix that ASAP :) Hope you found some music you love! cheers

While you're fixing UI please consider fixing Jamendo PRO search interface. Your database seems to have more information than you allow to filter by (I really need to filter by tags, e.g. I want something fast, dark and agressive). I spent days trying to find music for my game on multiple websites and I can definitely say that pretty much everyone's music search sucks. If you make yours awesome, you'll definitely stand out.

yes, good catch. we can improve the interface but also one of our main problems is the reliability of the current tag data, which is too low. We're considering a process with mechanical turk to fix it.

Anybody here with experience with mturk and this kind of metadata tasks ?

Just finally let us Jamendo users tag the music. Maybe weigh it less than the artists' own tags.

Uh-oh. No you didn't. The value for "C" is INCORRECT. ==> http://goo.gl/fpJsJ

Google Chrome blog has a post about this. Congrats to the Jamendo guys. http://chrome.blogspot.com/2010/12/x-g-chrom-3.html

Google's blog should be read carefully. There are errors in this. ==> http://goo.gl/fpJsJ

I can only view the cached version. Looks like he needs a mirror.

Or he needs to implement caching on his Wordpress site.
Almost every day there's a popular blog post that misses a large portion of its audience because the author wasn't caching. You'd think that more authors would plan for a traffic spike from aggregators, since that's sort of the holy grail of blogging.

It's hard to plan for a problem you've never had before. Most people expect it's just going to work.

Or that their host can handle the stress.

In the same vein as caching (and why caching is good) but the less DB activity each page request can generate, the better. Shared hosts seem to be very conservative with how many concurrent DB connections are allowed to be open at any one time, HostGator's limit is I believe 25. But persistent connections are usually available; PHP has mysql_pconnect and mysqli_connect('p:' . $host).

I've always had a lot of issues with mysql_pconnect in PHP. Hung connections, connection reuse causing data corruption, and various other random issues.
Generally using just mysql_connect is faster than using mysql_pconnect. On the server I admin for a large website with millions of hits a day we don't use mysql_pconnect.

> that's sort of the holy grail of blogging.
eh, I've lived through several spikes in the 6 years I've been blogging - the real holy grail is watching your search engine traffic increase bit by bit, month on month. I've been on digg's frontpage (back in 2007ish), linked to from xkcd's blog, I had a 20000 visit spike from reddit just this week. The spikes are fun, but they fade away after 2 or 3 days. The xkcd traffic lasted longer but that's because Randall didn't update his blog for weeks after the update that linked me.
But yes - even so, it's not nice to have your site die in the middle of a huge spike. It's just possible that that's the day Larry Page would have decided to tweet about you.

Reply on news.ycombinator.com