onecreativenerd

I'm testing it out and it seems pretty legit if you need a big, big key value store and don't want to wrangle redis or memcached shards.

Non-Java drivers are a bit behind, but nothing seems to have the same 95%tile latency, fast failover, SSD performance for actual read loads, and relatively easy scale out.

The export feature would be really awesome if it could handle larger documents!

Also, sharing a document with someone else seems to throw an exception when they try to sign in...

The background music is a little distracting. It would be nice if it were at the same volume the whole time.

Other than that, I wouldn't have minded if you talked faster, but I have an attention deficit. :)

Actually, you can put the monitor on the keyboard shelf, and the keyboard on the desk.

I had to do a little shimming of the shelf bracket, since the legs were not perfectly straight, but no wobbling here either.

The height of the shelf is perfect for sitting. You can put a laptop/monitor on it and the keyboard below for some variety.

I believe a haptic tablet experiment is in order...

Here's a first pass at auto-discovering topics in that text using LDA (Latent Dirichlet Allocation) in R: http://opani.com/ryan/sarah-palin-email-topics/148845806827/

The biggest deficiency now is the stopwords list is not working 100% so mundane topics are creeping in. I'll have limited time to play with it in the next few days, so feel free to make some cool discoveries. :)

thanks, that's good feedback. we are actually using ec2 (you can even use your own AMI image).

intended message: we give you the tools you already use and make the ux of scaling to bigger data and getting meaning out of it effortless.

we should also link to people's apps so you can see what's being done already before you sign up.

mouseover the image

Ah, but that might skew the results... maybe not too much, though since it's only manipulating the root document? However, cross-domain permission would be broken for javascript inside the site.

If you look at the bottom of ui.js, you can see I tried a brute-force framebuster-buster. It does notify the user that the site is trying to break out, but it also catches legitimate outgoing links. It needs to be polished more before I can release it.

This is issue #1 on github for the project, BTW.

I am only measuring the "onDOMContentLoaded" or jQuery's cross-browser equivalent. It's certainly possible for pages to load objects after that using javascript, which "cheats" a little for purposes of this comparison.

What really matters in practice is "time to first interaction" which is when the user thinks your site is useful. Next up on my feature list is having a button to indicate which pages seemed to load faster to you. The idea is you can send a link out to your friends and get their impressions, which is more valuable than the numbers you can get with an automated tool. :)

You might be right. I considered that, but it doesn't have as much of an impact factor on the user, which is really important for getting the word out about performance (why the site exists in the first place).

I probably doesn't matter too much as long as you have a ~3Mbps connection or higher (see Mike Belshe's "Effective Bandwidth of HTTP" graph on http://www.belshe.com/2010/05/24/more-bandwidth-doesnt-matte...) because the TCP transmission rate can't ramp up high enough on each object to saturate the connection. It's possible, however that the objects will interact with each other by causing differing queueing latency and possibly loss in the bottleneck router and by changing the order of scheduler events on the client OS. Also, it may be better subject the two pages to the same network conditions (at least in the last mile) while they're loading, esp. on mobile.

I believe Bing loads images asynchronously after the DOM content has loaded.

try moving the mouse around over the images and then look at your CPU graph