HNHacker News
TopNewBestAskShowJobs

jaylevitt

1,090 karma · joined August 31, 2011

submissionscomments
jaylevitt··on A 14-year-old could build 1998's Google using her Dad's credit card
AOL in the late 1990's, minus the dialup itself.

Content publishing: Weekend project. Rails, memcached and CloudFront and you're done.

IM and Buddy Lists: 1.5 million simultaneous users doing n^2 pub/sub-type distributed transactions.

Mail: 4,000 emails per second with live unsend and recipient read/unread status. I think PostgreSQL tops out in the millions of rows per second nowadays.

Web caching/acceleration: pick your favorite proxy solution and configure it.

Single sign-on: Form strategic partn-- Hey, you said technical challenge, not political.

jaylevitt··on JSON will be a core type in PostgreSQL 9.2
That's not going to help when you need to restrict your query with JSON, though; the select list is processed after the whole FROM clause, including JOINs, so you'd be producing the whole table and then throwing away whatever didn't match.

In general, a few core limitations of the current optimizer - lack of LATERAL key among them - make it impossible to create a PostgreSQL function that blends well with SQL, AFAICT.

jaylevitt··on Many Core processors: Everything You Know (about Parallel Programming) Is Wrong
It's not that the brain works "reliably enough"; any number of experiments will show that it doesn't.

It's more that one of the properties of the brain is that it believes itself to be reliable enough.

jaylevitt··on JSON will be a core type in PostgreSQL 9.2
hstore is all kinds of awesome, but one limitation is that the values are pure text. You can nest an hstore inside another, but only if you're smart enough to know that it SHOULD be treated as an hstore; as far as Postgres is concerned, it's text that happens to contain '=>'.

If JSON support means first-class type support in a nested object, that's a huge leap forward.

jaylevitt··on Automatically populating .ssh/known_hosts
I've been wrestling with this this week. I like the console-output idea, but doesn't console output take a good 5-10 minutes to show up?
jaylevitt··on Selling a 300-Year-Old Cello
For the love of God, do NOT let the buyer use Paypal.
jaylevitt··on Sublime Text 2 Beta released with Auto-complete and Improved UI
See elsewhere my post about ExpanDrive and SSH tunnels..
jaylevitt··on Sublime Text 2 Beta released with Auto-complete and Improved UI
Is SublimeCodeIntel working well again/yet? I tried it a few months ago on Mac and it caused major slowdowns all throughout ST2 - even just typing text was slow when it's installed. I've heard other such reports on the forums, and a co-developer had the same problem last week.
jaylevitt··on Sublime Text 2 Beta released with Auto-complete and Improved UI
OK, this isn't "decent". But: On a Mac, with a good connection to the remote machine, I've found the combination of

- ExpanDrive, which uses MacFUSE to mount remote volumes over SSH, but with lots of metadata caching smarts to make it fast and reliable

- My hacky remote textmate script, which gives you a "mate" command on the remote machine via ssh tunneling (https://github.com/jaylevitt/textmate_remote)

to work well with TextMate. I've been meaning to try it on ST2 but haven't yet; as long as ST2 doesn't do the horrible thing TextMate used to do and continuously rescan the project directory in a blocking thread, you'll be cool.

jaylevitt··on Why is C faster than Java: git vs JGit
Yep. The usual "Program X is faster in C than Java" gets a barrage of "That's because you know C better". Shawn is a performance-obsessed Java expert, Eclipse committer and longtime Google coder who works on JGit. If he says Java is slower than C at this, then Java is slower than C at this.

EDIT: but as wcoenen points out, this was written in 2009 and Java 1.7 does a better job with some of this.

jaylevitt··on List of Websites Monitored by Homeland Security published by Cryptome
Today I learned that Homeland Security thinks it's important to stay current on MySpace Videos, but not Bruce Schneier.
jaylevitt··on It’s time to place the web in safer hands
I think their use case is worse, but their mental model probably includes other TLDs - or at least they've probably GONE to other TLDs. Some (More? many? most?) people just prefer to type URLs into Google for some ungodly reason. I watched a junior web designer do it today, and she certainly knows what URLs are; it's just How You Do That. For you and me, it seems like an extra hop, but for her, the extra hop would be to realize "I already know this URL" and override her muscle memory.

Then again, I set up an alias called "gwssh" that goes to our EC2 gateway machine and then to the target machine, and more than once I've typed "gwssh gate" because I'm too "lazy" to type "ssh gate".

jaylevitt··on It’s time to place the web in safer hands
Why, oh why, does everyone assume that a plethora of TLDs implies that all those TLDs must contain a complete copy of .com?

I claim the problem with the first big expansion - .pro, .info, etc - is that it WASN'T wholesale. Yes, if half the Internet thinks that only .com exists, and you add .biz, every corporation has to go register .biz as well.

If you add a thousand real, active, well-used TLDs, they don't, simply because that is no longer how the Internet works. We already see the start of this with ccTLDs; Apple (a pretty aggressive trademark defender) does not own http://www.apple.fm, and nobody expects them to, and nobody is confused that they don't. I'm sure my grandmother sometimes confused Elvis Costello with Elvis Presley, but nobody thought the one was impersonating the other, because people have last names, and that's how it works, and that's normal.

Five years ago, the average American non-technologist's mental model of the Internet was "you type the company name and .com, except sometimes it's .net and I don't know why". Now it's probably a little more nuanced - but only a little, because .com is still Where Everything Is.

If you change that, you've changed that. Because you changed it. So it's not the same.

jaylevitt··on Wwwizer: Solve your naked domain problem. Put 'www' in front of your domain name
"triple-dub".
jaylevitt··on Microsoft patents "avoid ghetto" feature for GPS
For one highway and one non-highway, check out SPUIs: http://www.youtube.com/watch?v=mwpoPQ1SPJU
jaylevitt··on SOPA-Supporting News Outlets Aren't Covering SOPA
Ron Paul's been on the front page of the Washington Post web site nearly every day. And they're kinda mainstream.
jaylevitt··on How Yoga Can Wreck Your Body
Someone who knows more about anatomy than anyone I've ever met once said to me: "Half of all yoga poses are great if done correctly, and half of them should never be done by anyone." He mentioned the pigeon pose as a particular instance of a pose that may feel like it's "opening you up" but which, if you're doing it perfectly, is stretching something (hip ligaments) that don't need to - and shouldn't - be stretched.

EDIT: Downvoter, if you know otherwise, please educate me!

jaylevitt··on Why Don’t Smartphones Have A “Guest Mode”?
What kind of computer do you have that doesn't have a guest mode? Windows, Mac/UNIX and Linux have all been multi-user either from the beginning or for decades.
jaylevitt··on Bill Joy's greatest gift to man – the vi editor
rbanffy and ordinary are both right. "Baud" is the number of symbols per second; a symbol can be a specific frequency, or a phase shift, or Complex Electronic Magic. Up through 300 baud, each end used a tone for 0 and a tone for 1, and that was it. One bit = one symbol. (You could hear zeros and ones go by if you listened, and for Commodore owners, it was critical to be able to whistle the answer tone so you could trick the modem into hanging up - it auto-answered every call.)

So each baud carried one bit, and we got used to equating 300 baud with 300 bps. Then came 1200 bps modems, which sent two bits per symbol (four discrete tones, IIRC); they were technically 600 baud modems. It got more complex from there. And thus the confusion began; trying to tell people they had a 600 baud modem was about as useful as trying to tell them it's really called GNU/Linux.

jaylevitt··on Google demotes Chrome in search results over pay-for-post promo
Utter chaos, unless Google's anti-spam team is made up of humans, who could see through it.

See also: "If I change my name to Mark Zuckerberg, they can't sue me, because the judge will think I'm Mark Zuckerberg!"

jaylevitt··on Matt Cutts Responds on Google's Paid Link Campaign
From Google's statement:

"While Google did not authorize this campaign, and we can find no remaining violations of our webmaster guidelines, we believe Google should be held to a higher standard, so we have taken stricter action than we would against a typical site."

http://searchengineland.com/google-chrome-page-will-have-pag...

jaylevitt··on Matt Cutts Responds on Google's Paid Link Campaign
No, I think they demoted the URL, no matter what the search term; the initial searchengineland article was written while the demotion was still propagating, and "chrome" was still ranking. Now:

'In fact, the page no longer ranks for “chrome” or for “chrome browser,” either.'

http://searchengineland.com/google-chrome-page-will-have-pag...

jaylevitt··on Do you make these 5 database design mistakes?
Interesting.. how do you deal with ID collisions after rollover? That was our big problem, even with 8-byte IDs - the ID allocator had to know the Global Truth. Dealing with collisions as "exception, try the next one" was too expensive, though with faster hardware that may be moot.
jaylevitt··on Wikimedia Foundation Raises $20 Million From 1 Million+ Donors
I think you have it backwards. Wikimedia has a business model. Jimmy Wales looks solemnly at you, and then you give them money.

Why aren't you guys looking for ways to scale this up? If Jimmy Wales was solemnly looking at me when I went through the turnstyle, would fare evasion drop?[1]

Think big, man: Jimmy Wales as a Service. JWaaS.

[1] Actually, psych research suggests it would: http://www.newscientist.com/article/dn9424-big-brother-eyes-...

jaylevitt··on Do you make these 5 database design mistakes?
How do modern distributed OLTP systems deal with generating unique sequence numbers? Back in the day, this was a big problem, and I always thought that GUIDs would someday be a solution (though at the time, any string was too big/slow to be a primary key). Having one key-issuing server was a SPF; sharding the key ranges by server made it difficult to add new servers.
jaylevitt··on Why 13th Chords
Been out of Berklee a few years, but a few nuggets that might shed light:

In Jazz theory, the 7th scale degree is part of the chord. C7 isn't C (C+E+G) plus a B♭; C7 is its own flavor of C chord, C + E + G + B♭.

More generally, the ^3 and the ^7 are the two most important notes in the chord. The ^3 tells you its quality (minor, major) and the ^7 tells you its, uh, seventhness (dominant, diminished, maj7). It's common for players to leave off the 1 and the 5; they're low-information notes, and those fingers are better put to use playing the tensions, which are "chord extensions", the 9th, 11th and 13th. Critically, all three tensions are considered part of the chord, unlike Csus4, which means substitute a 4 for the 3, and don't play the 3.

C9, C11 and C13 can exist, but you won't often find C13 in particular like that. There's a whole set of memorizable rules to tell you which chord scales go with which chords performing which functions; if that C is a I chord in the measure, you're not likely to see a C11. You may see a C7(11), or C7(add11), which is IIRC the C chord plus an 11th note - but not the 9th. Then there's C7(#11), which you'll often find when C is the IV chord, 'cause the IV chord gets a Lydian scale.. but now I forget how that explains anything.

C+E+G+B♭+D+F+A can also be looked at as D-/C7, and often is. That whole "Upper Triad" polychord stuff is a whole semester to itself.

jaylevitt··on Godaddy No Longer Just Doesn’t Support But Actually “Opposes” SOPA
The Silver Lake connection may explain some of it, but GoDaddy has historically had a Rambo philosophy towards any content they deemed objectionable. When I was working on a startup that would use trademark law to shutdown phishers, 409s, botnets, and fake pharma, GoDaddy was eager to play in that arena, and was considered a good friend to law enforcement in general.

There's a difference between

"If you want to see our server logs, you're gonna need a warrant!" and

"If you want to see our server logs, you're gonna need a warrant; the Eastern District has pretty fast turnround and they know the landcape. Just bring that warrant back over here, and we'll get you the data you need."

jaylevitt··on Cracking WPA in 10 hours or less
I'm no crypto guy.. is there any conceivable situation in which their idea works? I mean, when is it actually more secure to say 'I'll let you know if you got the first half of the password right before you enter the second'?
jaylevitt··on Scientists' proposed calendar synchronizes dates with days
Obviously, the calendar ambiguity can be resolved by some sort of calendar NAT.
jaylevitt··on Cracking WPA in 10 hours or less
Good God, what a comedy of errors. First, that WPS even lets you get around a 256-bit key with a 10^8 PIN (like others, I thought WPS was pushbutton-only), but second, that this vulnerability brings the brute-force complexity down to 10^4 + 10^3, or 11,000 attempts: http://www.kb.cert.org/vuls/id/723755
← PreviousPage 8 of 12Next →