kiwibirdft

Hello FT peeps,

After using ITA Matrix for months, I noticed that there were a few occasions where ITA Matrix output an empty calendar instead of throwing "no flight result". I didn't pay much attention to this at first but since I am very keen on a recent construction that outputs an empty calendar, I wanted to know if theres any difference between the two.

My wild guess is that ITA knows the construction is possible but it runs out of computational resource to continue the search, hence presenting a "half-baked" result; whereas for the latter, ITA knows straight up that a construction does not obey to the fare rule and hence render it unconstructable.

lensovet

Sounds right to me.

in the vein of weird results, I’ve seen cases where searching with a wider scope somehow makes flights show up that should have been shown with a more narrow one.

for example searching for flights from EWR to ATH on SK would only show me itineraries with ridiculous layovers in CPH, but somehow adding JFK to the starting city (like “EWR,JFK”) made it find flights from EWR with quite reasonable layovers. This persisted even if I ran the original search with a MAXCONNECT criterion (which claimed that no flights could
be found).

that said, I was still able to find and purchase a ticket for a price that was probably 30% cheaper over what the airline or any other website would let me find, so I can’t complain too much. But it does make me wonder if I could have found an even better deal somehow (or on dates that worked better for me).

andydd

ITA Hacker I see a new bug in old matrix since yeterday. While searching for any city..the city picklist shows random cities..and also sometime the list appears after almost 4-5 seconds, when I have clicked the searched button. This was perfectly working fine before. Anybody else epxeriencing the same?

kiwibirdft

Yeap, kinda annoying that this happens. Like what you said, it requires hard coding to output the more optimised fares.

Recently, I found out that the devices in which we run our ITA matter too. For example, on my mobile, i could find SIN-LHR round trip on fare class O which is quite reasonable, but the same input when being keyed into my laptop, it gave 0 result. I then removed the booking class code in advanced control and it suggested me the least restrictive but expensive econ fare.

Not sure why device plays a part.. afaik it shouldn’t be affected by device specs since it’s an online query but perhaps anyone who knows how ITA matrix works under the hood can explain this :/

ChocolateFactory

Not sure how it works under the hood, but most likely there is simply some “randomness” in which results are returned first (for example, which server your request is routed to by the load balancer/what data that server has cached etc.), before the query times out and the results are returned.
I’d be very surprised if it had anything to do with the client device.

kiwibirdft

that makes sense, but I suppose each query, though from the same device, gets routed to a different server and on a random fashion no? I tried rerunning the same input multiple times but I kept getting the same result.

do you guys clear your cache after every query?

andydd

I think that unfortunately that Google is pulling the last ventilator support from beloved old matrix. It takes upto 10 seconds for any parameter to resolve..for eks sales city, currency src, destination pairs etc. Plus it's impossible to search for any side trip on old matrix anymore. It shows no results found contra to results shown on new matrix.

What a pity. I just hope that the new matrix GUI is optimized to be effective and efficient as the old one. The new GUI is almost all over the place and counter intuitive unlike other Google products.

andydd

ITA Hacker ITA Support
1. I found yet another bug on new ITA. If one enters a wrong airport code. The ITA displays the wrong code warning but doesn't say which city code is wrong..which would be immensely helpful. In any case, once one removes the codes, and actually re-enter the codes..the previous error warning seems to be sticky and won't resolve to real airport codes..even when the correct airport codes are entered. You see the behaviour in steps 1 & 2 to reproduce.

(Note that TUN and CAI are real airport codes that are easily resolved in destination field in step#2, but aren't resolved in source field due to sticky error warning).



2. I also wonder if something could be done about enormous unnecessary white space on new ITA compared to the old one.. May be you can introduce a new CSS to remove the white spaces.

Thanks

ExpatExp

Is Google Flights based on the same technology now? I have noticed the same with Google Flights recently.

kiwibirdft

Advanced control question for the pros:
Assuming that I am interested in a transatlantic trip and would like to look at all the possible constructions, entirely on BA metal.
I am okay with any amount of stopovers but I would like to have at least one of them on a 737.
With the current input, I can only see non-stop constructions. How should I write the extension code to accurately indicate my requirement?
Thanks.

ChocolateFactory

You can’t, as far as I know.

But you can do a little better at least.
First of all, you’ll want O:BA+ (not just BA+) to get BA metal (BA+ would also give you BA codeshare flights operated by AA).
Using O:BA+ will give you any flight operated by BA, no matter the flight number.

Then you could simply list all aircraft you are interested in like “aircraft t:737 t:777 t:350 t:787 t:380”. Since 737 typically aren’t served by 777’s, that should probably be pretty close to what you want. The important thing would be not listing t:320 etc.

Or you’d have to split the searches/use separate queries and aggregate the results yourself.

man4business

ITA Hacker ITA Support can I address an issue with the new interface, please?

Going from Round Trip to Multi City does not work as expected, I mean that if I have return flight which is A->B (dateX) and B->A (dateY) and I switch to Multi City I would expect exactly the same two legs present. This is not the case, I have to add the latter leg (B->A) manually (by Add flight), both the airport (A) and the date (dateY).

Another observation: is it really mandatory that we have those fancy styled decoration around airport acronyms? I would dream of going back to plain ASCII... at least perhaps a switch in the styles (if someone prefers this enriched but counterproductive presentation)?

anabolism

But there is no way that I've found to specify BA prime flights. Using "o:ba;ba" results in a "Bad route specification" error.

ChocolateFactory

Are you looking for -codeshare ?

anabolism

No, because that's an extension code, so it applies to the entire route. I might want an itinerary that includes any number of BA and AA flights, prime or codeshares, but want one or more specific legs to be an AA prime flight. For example, if "p" was available to designate a prime flight, a route code of "aa,ba+ dfw p:aa+".