Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

We can't send email more than 500 miles (2002)

web.mit.edu

Post reply on HN
Print view View original

131–140 of 141 posts

Re: We can't send email more than 500 miles (2002)

#131

Post by pontifier »

TurkishPoptart wrote: post #130
pontifier wrote: post #121

Earlier quoted context omitted.

This is absolutely one of the key formative stories that helped me to think about systems at light speed scales. I'm currently at the very early stages of building a science museum and will eventually try to incorporate this story into an exhibit about light speed. This along with "Nanoseconds", foot long pieces of wire like what Grace Hopper handed out, can truly help to bring this topic to life. I'm also attempting…

I actually don't understand the last part of the OP story - can someone explain it?

The sending mail server was configured with a zero timeout for connections. If it didn't IMMEDIATELY get a response from the destination mail server it would fail. This immediate failure took 3 milliseconds, a long enough time that some servers could actually respond back before that happened... but if the server was too far away (more than 500 miles) the connection would fail before the first packet could even get there due to the finite speed of light.

Re: We can't send email more than 500 miles (2002)

#132

Post by photonic277 »

abhgh wrote: post #128

I used to collect these kind of stories: When I flush my toilet my computer reboots: http://www.techtales.com/tftechs.php?m=199712#66 (the first story on the page) If I buy vanilla ice-cream my car wont start: http://www.netscrap.com/netscrap_detail.cfm?scrap_id=501 A specific cargo routing crashes system: https://www.jakepoz.com/debugging-behind-the-iron-curtain/ Tape-drive failure only within large print jobs: http…

In my intern days some time around 10 years ago, a PI at the NASA GRC facility told me about a problem of this flavor an old grad student of his had.

The guy was working on an optical sensor in a light-tight lab. Every morning, he came in, calibrated the sensor, and performed measurements. All morning, it held calibration with negligible drift. But when he came back from lunch, each time, the calibration had drifted off.

Could it be related to the time of day? He tried taking his lunch an hour earlier and an hour later. Each time, the calibration was rock solid until right after lunch.

In spite of protocol, he tried eating lunch in the lab, no one else in or out. Before lunch: good calibration. After lunch: bad calibration.

He tried not eating lunch at all. That day, the calibration held all day.

How could an optical sensor have any concept of whether its user had eaten lunch? It turned out, it only had to do with the lunch box. The sensor was fiber coupled, and it was sensitive to changes in transmission losses generated by changes to local radii of the patch chord. Every morning, the grad student set his lunch box down on the lab bench, nudging the fiber into some path. After eating, he’d replace his lunch box on the bench, nudging the fiber into a different path.

After that, the fiber was secured with fixed conduit, and lunch boxes no longer entered the lab.

Re: We can't send email more than 500 miles (2002)

#133

Post by reaperducer »

staycoolboy wrote: post #58

The one thing that throws a wrench in this story for me: Lattes. Latte's in 1994? In North Carolina? No way. Maybe on the West Coast, but I moved to Cali in 1989 and they were a rarity until the mid-late 90's. There were only 425 starbucks in the US in 1994 (from their site). The "fancy coffee" craze was just a blip on the radar in the mid 90's but gaining momentum. ;-)

Latte's in 1994? In North Carolina?

We drank lattes in Louisiana in the 80's. Time to upgrade your stereotypes.

Re: We can't send email more than 500 miles (2002)

#134

Post by petercooper »

wuster wrote: post #59

Best part of reading this is coming away having learned the existence of units the CLI. How did I spend 20 years on the shell and not have needed or discovered this?

Though sadly millilightseconds is not supported on macOS, at least, so you have to go:

    3 millilightyears / 365 / 86400
Of course, round 365 to whatever average number of days you believe in :-)

Re: We can't send email more than 500 miles (2002)

#136

Post by chriselles »

pontifier wrote: post #131
TurkishPoptart wrote: post #130

Earlier quoted context omitted.

I actually don't understand the last part of the OP story - can someone explain it?

The sending mail server was configured with a zero timeout for connections. If it didn't IMMEDIATELY get a response from the destination mail server it would fail. This immediate failure took 3 milliseconds, a long enough time that some servers could actually respond back before that happened... but if the server was too far away (more than 500 miles) the connection would fail before the first packet could even get t…

As a tech/coding newb I was assuming it related to speed of signal travel to approximate 500 ish miles.

Thanks so much for confirming it.

I‘m going to share this one with my sons who will appreciate the humour.

Re: We can't send email more than 500 miles (2002)

#137

Post by m000 »

cranium wrote: post #77
wuster wrote: post #59

Best part of reading this is coming away having learned the existence of units the CLI. How did I spend 20 years on the shell and not have needed or discovered this?

One thing I got bitten by was the handling of Fahrenheit/Celsius, because it's a non-linear conversion between the two. When you ask to convert `10 degC` to `degF` you get 18, which is the delta of ºF corresponding to increment of 10ºC. To get the absolute temperature, you have to ask to convert `tempC(10)` to `tempF` which is 50, as expected. https://www.gnu.org/software/units/

FWIW, units on macOS (not GNU) handles conversion of `10 degC` to `degF` correctly, although it dates back to 1993.

It seems that GNU units at some point added support for several non-linear units, which may have prompted them to rethink their syntax.

Re: We can't send email more than 500 miles (2002)

#138

Post by MasterYoda »

Haha, this was interesting. I posted the same story with more or less the same title a couple of month ago [1], and no one saw it and no comments. This time it got almost 1000 points and a lots of comments. What I think is interesting is how the same thing can get so different traction. Wonder what factors it is that makes a thing get traction and not?

[1] https://news.ycombinator.com/item?id=22164691

Re: We can't send email more than 500 miles (2002)

#139

Post by dang »

franky47 wrote: post #102
techdragon wrote: post #98

Earlier quoted context omitted.

Yes. Seriously... I’m willing to pay for the shipping and everything. I’ve told this story countless times to people over the years. This and the “OpenOffice can’t print on Tuesday” bug are two of my favourite troubleshooting stories.

HN thread for the OpenOffice story: https://news.ycombinator.com/item?id=8171956

Also https://news.ycombinator.com/item?id=11717010

and https://news.ycombinator.com/item?id=22157430

Re: We can't send email more than 500 miles (2002)

#140

Post by tzs »

comicjk wrote: post #96
cranium wrote: post #77

Earlier quoted context omitted.

One thing I got bitten by was the handling of Fahrenheit/Celsius, because it's a non-linear conversion between the two. When you ask to convert `10 degC` to `degF` you get 18, which is the delta of ºF corresponding to increment of 10ºC. To get the absolute temperature, you have to ask to convert `tempC(10)` to `tempF` which is 50, as expected. https://www.gnu.org/software/units/

"Non-linear" threw me off for a second - I almost never see the mathematically correct definition of linear in computer science spaces. For anyone wondering, Celsius to Fahrenheit is an affine transform, technically not linear, because you have to add an offset, not just multiply.

On the other hand, an equation of the form y = a x + b is a linear equation. If you have Celsius and want Fahrenheit you accomplish that by applying a linear equation (F = 1.8 C + 32), so I certainly can't fault people for saying that the transformation they are doing is linear.

I wonder what people would say for something using an equation of the form y = a x^2 + b x + c to transform something? I can't say that I've heard anyone talk of quadratic transformations. On the other hand, I can't think of ever transforming anything with a quadratic equation, so never had the need t speak of it.

(Also, he called it a linear conversion, not a linear transformation).

Post reply on HN