News for hackers. Stuff that matters.
Re:Ryanair, Hamiltonian Cycles, and using graph theory to fi... by kasparsklavins on Monday November 26, 2018 @09:05AM (#18533389)
Why is there no interface that allows searching for flights without specifying exact dates? All I want to do is fly from A to B to A. Maybe stay a week, maybe two. I'm flexible - it doesnt have to be exactly N days starting on the M'th.

Even services that "allegedly" advertise they have that, still require manually clicking every day combination.

I'd fly more often if it was easy. The few times I tried, it would be just faster to drive than go through all possible 900+ day combinations.

Re:Ryanair, Hamiltonian Cycles, and using graph theory to fi... by empath75 on Monday November 26, 2018 @09:53AM (#18533702)
Skyscanner has this. You don't even need to specify a city:

These are the results for 'cheapest month' and 'everywhere' from DC

https://www.skyscanner.com/transport/flights-from/wasa/?adul...

Re:Ryanair, Hamiltonian Cycles, and using graph theory to fi... by s_dev on Monday November 26, 2018 @12:30PM (#18535054)
+1 for Fuzzy searching with Sky Scanner.
Re:Ryanair, Hamiltonian Cycles, and using graph theory to fi... by crazygringo on Monday November 26, 2018 @09:16AM (#18533463)
Have you seen a site that does that really well, UX-wise?

I'd assume it's two reasons: first that it's hard to get the UX right. There are going to wind up being so many combinations of possibilities that I'm not sure how you'd even sort them in a meaningful way. (E.g. it's unlikely price from lowest to highest will be the most meaningful, the lowest price alone might wind up having 100 different date combinations, so figuring out how to present the options in a meaningful way might just be something we haven't figured out yet.)

But second, it might just be too computationally expensive. You see how slow sites already are just for searching a single date, because there are so many possible combinations of connections to consider, and so many fare rules to calculate prices out of. Now a 2-week range for both start and end results in taking almost 200 times longer. It just might not be feasible, or worth the programming effort. I remember a blog post a long time ago from Kayak (I think?) talking about the insane effort that was required to cache flight fares just so show simple fare comparisons across a few days... and it only worked for the flights that were more commonly searched for.

Re:Ryanair, Hamiltonian Cycles, and using graph theory to fi... by RussianCow on Monday November 26, 2018 @02:37PM (#18536155)
adioso.com (no longer maintained as far as I can tell, but still online) does a really great job of this, UX-wise. It lets you create searches like "departing anywhere between February 10 and March 20 for about 2 weeks", and allows you to filter destinations by continent/region, which gives you a ton of flexibility. They then allow you to choose a destination (showing what I assume is the lowest price found per destination), and then let you sort the actual flights by cheapest, fastest, and "best" (a combination of price, duration, and number of stops).

I think the computational cost is a far bigger problem for these kinds of services than the UX.

Re:Ryanair, Hamiltonian Cycles, and using graph theory to fi... by rb808 on Monday November 26, 2018 @09:48AM (#18533668)
I've seen a few airlines have grids where the outbound and return flights are shown on each access, which I think is great. I'm thinking of Air New Zealand as the last airline I saw it so maybe it benefits from a smaller network as you say.
Re:Ryanair, Hamiltonian Cycles, and using graph theory to fi... by hoschicz on Tuesday November 27, 2018 @04:42AM (#18540922)
Kiwi.com does this really well.
Re:Ryanair, Hamiltonian Cycles, and using graph theory to fi... by thetest3r on Monday November 26, 2018 @09:07AM (#18533411)
Google Flights will render a whole calendar of prices.
Re:Ryanair, Hamiltonian Cycles, and using graph theory to fi... by rockostrich on Monday November 26, 2018 @10:16AM (#18533849)
Hopper does this with our "Flex dates" feature. You can specify a date (up to a year out I think) and stay (anywhere from 4 days to 2 weeks) window for when you want to fly as well as regions you're interested in traveling to, instead of just 1 destination (although we also allow you to select just 1 destination).

It was sort of hidden for a while because we were testing other features, but it's back on the main search screen now.

Disclaimer: I work for Hopper

Edit: Forgot to mention this feature is only available on iOS at the moment...

Re:Ryanair, Hamiltonian Cycles, and using graph theory to fi... by jirisykora83 on Friday November 30, 2018 @04:50AM (#18567944)
I think it´s because most of the UX is designed a long time ago. The pricing wasn't too dynamic and search engines/airlines were more focused on business travelers (from which they have more money) and they usually don´t care too much, how much it cost and when they must / want to fly.

We are building our search engine (focus on EU low-cost mainly) which allow what you looking for (even if you want to search this) for example: A -> B [stay 2-6] -> C [stay 10-12] -> A

Re:Ryanair, Hamiltonian Cycles, and using graph theory to fi... by urtrs on Monday November 26, 2018 @09:42AM (#18533630)
Maybe https://adioso.com is close to what you want. I found it recently but I have not tried it yet
Re:Ryanair, Hamiltonian Cycles, and using graph theory to fi... by amyjess on Monday November 26, 2018 @12:58PM (#18535340)
If you're flying domestically, check out Southwest's low fare calendar.
Re:Ryanair, Hamiltonian Cycles, and using graph theory to fi... by findjashua on Monday November 26, 2018 @10:30AM (#18533972)
Southwest provides this through their “low fare calendar” feature.
Re:Ryanair, Hamiltonian Cycles, and using graph theory to fi... by m__ on Monday November 26, 2018 @09:35AM (#18533584)
kiwi.com allows searching for time of stay, e. g. between 2-10 days, but not providing any exact dates.