ntoshev

Tornado devs recommend using blocking mysql drivers and disk I/O. These operations have relatively low and predictable latencies. You can run more Tornado processes than the number of cores is if CPU is underutilized.

What are you referring to?

Mount this space as a filesystem:

http://code.google.com/p/google-docs-fs/

We all know services like Gmail can be subpoenaed by court to provide information. Google is big enough for this to happen to them on a daily basis, so it's very likely they have a an automated procedure. For all we know, Schneier may be talking about this.

Do you have any link describing this backdoor and the procedures to use it (grandparent's claim "without judicial oversight" is the critical bit here)? Schneier just mentions it without reference.

I didn't exactly mean a Redis server implemented in Python instead of C... It could be that, but speaking application-specific DSL instead of generic data structure DSL.

Redis is i/o bound, while this approach eliminates i/o and copying data from Redis to the client ;)

Not sure about memory efficiency. You can use __slots__ in Python or custom data structures (e.g. blist) if you have specific needs.

I've had an alternative Redis-like design for a while... What if you have a Python process that keeps the data in dicts and lists and Python objects as usual. These data structures get persisted to disk by forking and pickling the data as a snapshot, while the main process continues to serve requests. For a small server, the Python process can be the single-threaded web server itself (e.g. using Tornado web server).

I have this pet hypothesis that matching delicious tags with actual words found on the web pages will make for a killer query expansion algorithm...

This result suggests that a limit of +4 displayed points would be equivalent to the current situation, while leaving more visual guidance about how valuable a comment is perceived to be. If you chose to do that, maybe you should color-code it instead of displaying a number, though.

FWIW: I like the comments with hidden scores better than before.

It doesn't quite work, because Steve says "you need to be more open to fixing annoyances" and Rich responds "Look what a mess would it be if we fixed everyone's annoyance!". There is a big continuum between the place Clojure is now and the hypothetical absurd described by Rich.

I don't actually write object oriented (class-based or prototypal) code in Javascript, I write functional code instead and it seems to work well. Am I missing something? Any examples of "actual applications that are best modeled with prototype chains"?

I was also wondering recently why illiteracy carries social stigma, while innumeracy does not. The article asserts everyone just considers math ability to be innate, but I'm not sure this is a good explanation.

Our daily life seems to rely on literacy much more than it relies on numeracy. You can only know how to add and subtract (mostly money) and you will still function all right in your daily life and a great variety of jobs. Not so much if you have difficulty reading. This seems to be slowly changing in the future, but if you consider the learning curve, mathematics has a much steeper one compared to reading and writing well.

Does anyone have a better explanation?

> Stock price only starts to matter when a company needs access to more cash

I used to think that too, but it's not true. Stock price matters for the existing shareholders, including founders and employees, who may need to cash out for personal reasons. The stock price is an instrument to align the interests of many.

Scaling a website and scaling a search engine are two very different things. Python websites are OK, Python search engines... well: http://whoosh.ca/blog/fools_errand

Friendster. Maybe there are lot's of others who just had technically competent competitors, so we never knew about them in the first place.

Interestingly, Chris Dixon's own Hunch has a pretty good potential for SEO with their long tail of multiple answers questions (teach-hunch-about-you), but they do not exploit it. These are user-generated questions and answers, currently hidden from Google in Javascript. Text in the form of questions typically ranks well because this is what users often type in the search engines, and very little text on the web is phrased as questions.

Python:

max((s[start:end] for start in xrange(len(s)) for end in xrange(start, len(s)) if s[start:end]==s[start:end][::-1]), key=len)

What is the use of this question without context? You could check the ratings/stats on TED.com and pick what is most popular. Quora is slightly different demographics than general TED audience, but is this minor difference worth asking such a question? I don't think so.

It seems the canvas will be hardware-accelerated.

http://developer.android.com/sdk/android-3.0-highlights.html

Hardware-accelerated 2D graphics

Android 3.0 offers a new hardware-accelerated OpenGL renderer that gives a performance boost to many common graphics operations for applications running in the Android framework. When the renderer is enabled, most operations in Canvas, Paint, Xfermode, ColorFilter, Shader, and Camera are accelerated. Developers can control how hardware-acceleration is applied at every level, from enabling it globally in an application to enabling it in specific Activities and Views inside the application.

node.js is asynchronous from the ground up, Tornado is generally synchronous with the exception of http client and server. So, Tornado is less pure, but you can use any of the Python libraries as they are. Node.js is pure, even disk i/o is asynchronous, but there are much fewer libraries.

Tornado also wraps Facebook API and has authentication with Facebook, Google, Twitter and everyone else.

I don't know what Greplin is doing, but indexing alone doesn't imply they cache your data somewhere. Although the index contains pieces of your text, you can't reconstruct it from the inverted index alone.

That's too far from the bike concept and way too expensive.

I may try electrifying my bike, a local guy does this and it looks reasonable. You can see the pictures here:

http://prodavalnik.com/sglobyavam-elektricheski-velosipedi-p...

For the electro-motor, inverter, controls and batteries you pay:

1. 250W motor 36V+ lead-acid battery (3 х 12V 9Ah)- 300 €

2. 250W motor 36V+ lead-acid battery (3 х 12V 14Ah)- 323 €

3. 250W motor 36V+ li-ion battery (36V 10Ah)- 524 €

4. 250W motor 36V+ lithium iron phosphate battery (36V 12Ah)- 633 €

4. 250W motor 36V+ lithium iron phosphate battery (36V 15Ah)- 790 €

5. 500W motor 36V+ lead-acid battery (3 бр.х 12V 9Ah)- 338 €

6. 500W motor 36V+ lead-acid battery (3 бр.х 12V 14Ah)-363 €

7. 500W motor 36V+ lithium iron phosphate battery (36V 12Ah)- 674 €

8. 500W motor 36V+ lithium iron phosphate battery (36V 15Ah)- 830 €

9. 800W motor 48V+ lithium iron phosphate battery (48V 15Ah)- 980 €

10. 1000W motor 48V+ lithium iron phosphate battery (48V 15Ah)- 1025 €

Range is 30-80 km and you can always pedal more. I think this product could be a hit, with a bit better design and manufactured at scale.

It will work better for Instapaper. I use it regardless of the site formatting, maybe biased towards the longer pieces, which is a good thing.

Why do you care if you have mental illness or not? Either way, you have a bit different perspective on the world from everyone else, drugs could make you feel better, etc. Mental illness is just a label.

Try talking to your close friends about how the world is and how you'd like it to be (no need to get into details about leaving it). Maybe you'll find you can do something about it, albeit small.

> And does that mean that you store those pages in your database and use your own ranking algorithm, or are you using their ranked results for a query and then rearranging them on the fly?

I have looked at those search APIs and they don't let you write your own ranking formula. They just give you back 10-20-30 top results. There is no info on term frequency, inverted document frequency, pagerank or any other component of a decent ranking formula. So you can rank yourself the pages from your own crawls, but you can't merge this nicely with the API results.

If I were Garbiel, I would merge them in some crude way basically assuming my results are better, and add those URLs to the list for my crawler. The top URLs for all one and two word queries aren't that many.

People are not used to tracking, but they will become. Web companies aren't going to stop doing it, as it is giving up enormous competitive advantage and quality of service.

DDG's "not tracking you" for example means Gabriel can never offer you personalized search based on your search history (he wouldn't be able to anyway as it's not in the APIs and it is impossible to be implemented properly on top of existing APIs). This is a tangible hit on the quality of your search results right there.

SEO is supposed to be StackOverflow's core competency. They are completely aware most people end up on their site via Google. The search on their own site sucks.

The reason Q&A sites are so visible is that people tend to type questions in their search engines, so Q&A sites are a good match to those.

You can redefine anybody's business to look like he is competing with anyone else. The point is, those Bing and Yahoo ads still run on Google, so maybe there is some other reason for banning some affiliates, eh?

So a Java applet can listen on a port? Interesting, although not very practical.

Adobe Flash is also supposed to support some form of p2p connections, but I don't know anything about it.