Y
HN Search
Hacker News Search
new
|
comments
|
top
|
jobs
yid
searching PlanetScale…
1.
▲
2.
▲
3.
▲
4.
▲
5.
▲
6.
▲
53 ms
·
511.
▲
by
yid
15y ago
Wouldn't it make more sense to replace gnome/kde with gecko? Isn't that essentially what the win8 demo was all about? So many questions, so little cloud...
512.
▲
by
yid
15y ago
Perhaps not on modern on browsers. In Chromium on Linux, typing this in the dev console: x=new Date().getTime(); for(j=0;j<10000;j++); y=new Date().getTime(); for(j=0;j<10000;j++); z=new Date().getTime(); alert(x+' '+y+' '+z)
513.
▲
by
yid
15y ago
I agree with everything you said, except the last sentence. Circles is a competent product and FB is taking it seriously, but FB will only really have competition when Circles has comparable adoption (since nothing else really matters for
514.
▲
by
yid
15y ago
> * the timing functions you get from the browser are way too low resolution.* The JS time function returns epoch milliseconds, not seconds. At the scale of network latency (hundreds if not thousands of milliseconds), the lower order bi
515.
▲
by
yid
15y ago
"Purple air sounds like rage." Take some acid, and that sentence might make perfect sense.
516.
▲
by
yid
15y ago
It's going to be from ThePast@futuremessage.mailgun.org
517.
▲
by
yid
15y ago
I'm very familiar with MT from my PhD research :) It's a nugget of gold hidden in a single .c file. I asked the question in relation to the LC generators in stdlib, where it seemed like reseeding from the (probably MT? I believe BSD uses Ya
518.
▲
by
yid
15y ago
Just curious...why exactly would you say reseeding adds bias? AFAIK, srand in Perl on Linux uses /dev/urandom, which uses at least some bits from the hardware entropy pool. Swapped out rand() for MT in any case... :)
519.
▲
by
yid
15y ago
Interesting, I didn't know Perl's rand() was that terrible. Hopefully, the auto-seeding using srand will add some entropy. Also, since I'm choosing an epoch second, the times should hopefully be uniformly distributed at the scale of days.
520.
▲
by
yid
15y ago
The delivery script drops them after sending. :) I really have no interest in the email addresses or the messages -- this was just something I wanted for myself, and something that is intended to be write-run-forget.
521.
▲
by
yid
15y ago
I don't really have a user demographic in mind, this was just a fun project that was easy to set up. I'm using mailgun to deliver the mails, or rather I will use mailgun in about a month.
522.
▲
by
yid
15y ago
The exact second is picked uniformly at random using an approximation of 30 days per month. The code is: my $sendAt = time + int(rand()*86400*30*5) + 86400*30;
523.
▲
by
yid
15y ago
Lets say you get cheapo hosting @ $5/month. 40 years = $2400, which isn't a terrible investment. Although you'd probably want a VPS where you can ensure that the OS setup doesn't change over 40 years...
524.
▲
by
yid
15y ago
The response has been amazing! The DB currently has 153 messages, 3 hours after launch... :)
525.
▲
by
yid
15y ago
Thanks, I'll be sure to enable replying to the originating email address... :)
526.
▲
by
yid
15y ago
Just a side project, as far as pushing forward, I'm pretty done :)
527.
▲
by
yid
15y ago
Thank you SO much for the feedback...working on some of those changes right now. EDIT: changes are now live, thanks again!
528.
▲
by
yid
15y ago
I just set it up, and I can assure you that it's reliable (uses Mailgun for delivery, which is top-notch). Very interesting to look at futureme.org, I guess the difference here is that the time is chosen randomly for you, which allows an el
529.
▲
by
yid
15y ago
I set up the site as a curiosity (and to use it myself), and I can assure you that it works. It will, however, be at least 1 month till you get that citation (but more likely several more months). I use it to record startup and research ide
530.
▲
Show HN: Enter a message, delivery at a random time up to 6 months from now
(futuremessage.me)
66 points
by
yid
15y ago
|
52 comments
531.
▲
by
yid
15y ago
No, definitely not the whole city. I can usually cache strategic bits between wifi hotspots though, which gets me by.
532.
▲
by
yid
15y ago
I'm not sure what you're appealing to. Real-world analogies don't always transfer readily to digital law. Even in the words of the ageing CFAA, "exceeding authorization" to steal commercial information is most certainly a punishable crime.
533.
▲
by
yid
15y ago
It isn't breaking and entering if someone repeatedly trespasses somewhere (say, banned from a store) even if they change their clothes to avoid detection. It might not be breaking & entering, but it's still trespassing, which is a cri
534.
▲
by
yid
15y ago
> Does anyone think it's odd that an Acer laptop could write these files to disk faster than JSTOR could serve them? Why would that be odd? SATA = 3 Gbps throughput with minimal overhead, Ethernet = 1 Gbps with lots of overhead (IP head
535.
▲
by
yid
15y ago
"normalize" is a relative term. Does CPAN normalize Perl? Does STL normalize C++? The prevalence of a suite of libraries and tools is a good thing.
536.
▲
by
yid
15y ago
Re: closure compiler, not to mention dead code removal, type checking, conditional compilation via @define, function inlining, and a configurable flood of useful warning and error messages. I, too, dumped Crockford for Closure on a large we
537.
▲
by
yid
15y ago
Allow me some meta, meta-surprise that your comment isn't downvoted.
538.
▲
by
yid
15y ago
On an iphone at least, I cache the maps for the area I'm going to be in, then switch to airplane mode with location services turned on. Free, international GPS.
539.
▲
by
yid
15y ago
Um, what's the difference? Targeted online advertising involves enough research talent and money to make it not totally unlike an "increasingly convoluted financial instrument".
540.
▲
by
yid
15y ago
If anyone ever wondered what the phrase "cargo cult science" referred to, this is a prime example. They're going through all the motions, but sadly their understanding of the universe is gratuitously flawed.
More ›