4 ms·

Double Saw Bye (2009)

50 points by concerto 9y ago

falcolas 9y ago
I love this kind of problem solving. Doing this safely is probably a time consuming pain, yet less time consuming and dangerous than steaming several miles backwards.
woodruffw 9y ago
Problems like these bear a lot of similarity to CS brain teasers, such as exchanging N items without using N + 1 slots. [1] I wonder if that contributes to why model railroad clubs (like the TMRC) and early computer hobbyist groups intermingled so easily. [1]: https://en.wikipedia.org/wiki/XOR_swap_algorithm https://en.wikipedia.org/wiki/XOR_swap_algorithm
- mannykannot 9y ago
  In the first section of his 1984 book 'Hackers', Stephen Levy wrote about the Signals and Power Subcommittee of the MIT Tech Model Railroad Club, which included quite a few computing pioneers among its ranks. Levy mentions that one of the skills of the subcommittee was in disrupting any meeting through creative use of Robert's Rules of Order. https://en.wikipedia.org/wiki/Tech_Model_Railroad_Club#System_layout https://en.wikipedia.org/wiki/Tech_Model_Railroad_Club#Syste...
thriftwy 9y ago
I don't see why you won't have maximum train length for every part of track & check that beforehand while routing. Reminds me of the infamous MTU networking problems.
- PLenz 9y ago
  While through manifest trains are stable length locals pick up and set off cars along the way - and historically there was no reliable way you'll know exactly how many cars you'll switching in and out. Even today you can see conditions change after a local leaves a yard and before it reaches say an interchange between two railroads or a major shipper.
  - thriftwy 9y ago
    Why don't they make an app to track trains' length? Indeed I have hard time imagining an unaccounted car running on public rail.
    - falcolas 9y ago
      Human error. These kinds of things happen all the time - packages going through UPS/FedEx get lost, containers on ships are lost (i.e. put in the wrong stack or fall off into the sea), containers in ports are lost/hidden... There's a lot of rail cars in the US by itself. And while they are already well tracked, mistakes happen.
- Someone 9y ago
  This almost doubles rail capacity. There is an easier way to do that (run a single train of double the length, if needed with two locomotives) but it introduces larger latency.
mannykannot 9y ago
I did not get any hits on 'single saw bye' - Is that the case where a siding is connected to the main line at only one end? When I get a chance, I will see if I can figure out that one (I guess there are two variants of the single-junction siding, one with one and one with two switches.) Then there is the question of the minimum-length siding for arbitrarily-long trains. There seem to be some similarities to the Towers of Hanoi problem.

50 points by concerto 9y ago

I love this kind of problem solving. Doing this safely is probably a time consuming pain, yet less time consuming and dangerous than steaming several miles backwards.

woodruffw 9y ago

Problems like these bear a lot of similarity to CS brain teasers, such as exchanging N items without using N + 1 slots. [1] I wonder if that contributes to why model railroad clubs (like the TMRC) and early computer hobbyist groups intermingled so easily. [1]: https://en.wikipedia.org/wiki/XOR_swap_algorithm https://en.wikipedia.org/wiki/XOR_swap_algorithm

mannykannot 9y ago
In the first section of his 1984 book 'Hackers', Stephen Levy wrote about the Signals and Power Subcommittee of the MIT Tech Model Railroad Club, which included quite a few computing pioneers among its ranks. Levy mentions that one of the skills of the subcommittee was in disrupting any meeting through creative use of Robert's Rules of Order. https://en.wikipedia.org/wiki/Tech_Model_Railroad_Club#System_layout https://en.wikipedia.org/wiki/Tech_Model_Railroad_Club#Syste...

thriftwy 9y ago

I don't see why you won't have maximum train length for every part of track & check that beforehand while routing. Reminds me of the infamous MTU networking problems.

PLenz 9y ago
While through manifest trains are stable length locals pick up and set off cars along the way - and historically there was no reliable way you'll know exactly how many cars you'll switching in and out. Even today you can see conditions change after a local leaves a yard and before it reaches say an interchange between two railroads or a major shipper.
- thriftwy 9y ago
  Why don't they make an app to track trains' length? Indeed I have hard time imagining an unaccounted car running on public rail.
  - falcolas 9y ago
    Human error. These kinds of things happen all the time - packages going through UPS/FedEx get lost, containers on ships are lost (i.e. put in the wrong stack or fall off into the sea), containers in ports are lost/hidden... There's a lot of rail cars in the US by itself. And while they are already well tracked, mistakes happen.
Someone 9y ago
This almost doubles rail capacity. There is an easier way to do that (run a single train of double the length, if needed with two locomotives) but it introduces larger latency.

mannykannot 9y ago

I did not get any hits on 'single saw bye' - Is that the case where a siding is connected to the main line at only one end? When I get a chance, I will see if I can figure out that one (I guess there are two variants of the single-junction siding, one with one and one with two switches.) Then there is the question of the minimum-length siding for arbitrarily-long trains. There seem to be some similarities to the Towers of Hanoi problem.