iRail / hyperRail

The repo for the iRail.be webapp
https://iRail.be
Creative Commons Zero v1.0 Universal
46 stars 17 forks source link

Loading failure #39

Closed brechtvdv closed 10 years ago

brechtvdv commented 10 years ago

When I try to search for my train from Ghent to Antwerp in the morning, it keeps loading without results.

https://irail.be/route?to=008821006&from=008892007&date=070714&time=0850&timeSel=arrive&auto=true

nicoverbruggen commented 10 years ago

Looks like the website was updated and something broke. Doesn't work on my end either when planning a trip between Mechelen and Ghent.

Might be related to the Javascript merge (w/ grunt) with jQuery bugging out? Seeing as getting array length and reversing arrays doesn't work, which is jQuery.

pietercolpaert commented 10 years ago

Looks like an appcache problem on the one hand and an api issue in the other. Try opening a private or incognito window

pietercolpaert commented 10 years ago

Looks like the problem is really an API problem. The interface we scrape returns an awkward error:

<?xml version="1.0" encoding="iso-8859-1"?>
<ResC xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" 
   xsi:noNamespaceSchemaLocation="http://hafassrv.hacon.de/xml/hafasXMLInterface.xsd"
   ver="1.1" prod="String" lang="EN">
     <Err code="F2" text="Spool: Error writing the spoolfile." level="E"/>
</ResC>
pietercolpaert commented 10 years ago

I'm going to stop fiddling with this for today. If the problem is not resolved at the side of NMBS tomorrow morning, I'm going to write a new scraper tomorrow on a different interface (their new website runs on an updated version of their software, but the old one is still online).

Here's how you can test this:

## We have 2 interfaces to scrape from:
#http://www.belgianrail.be/jp/sncb-nmbs-routeplanner/extxml.exe
#http://hari.b-rail.be/Hafas/bin/extxml.exe
curl -XPOST http://hari.b-rail.be/Hafas/bin/extxml.exe http://www.belgianrail.be/jp/sncb-nmbs-routeplanner/extxml.exe -d '<?xml version="1.0 encoding="iso-8859-1"?>
<ReqC ver="1.1" prod="iRail" lang="EN">
<ConReq>
<Start min="0">
<Station externalId="008814001#80" distance="0">
</Station>
<Prod prod="1111111000000000">
</Prod>
</Start>
<Dest min="0">
<Station externalId="008885704#80" distance="0">
</Station>
</Dest>
<Via>
</Via>
<ReqT time="16:34" date="20140629" a="0">
</ReqT>
<RFlags b="0" f="6">
</RFlags>
<GISParameters>
<Front>
</Front>
<Back>
</Back>
</GISParameters>
</ConReq>
</ReqC>
'
pietercolpaert commented 10 years ago

Issue still not resolved by NMBS. Will work on a new scraper. Estimated time of fix: 3pm

pietercolpaert commented 10 years ago

And for some reason it magically works again. I'm going to work on an alternative scraper nonetheless :)

pietercolpaert commented 10 years ago

And reintroduced... Oh my God...

ChristopheVersieux commented 10 years ago

Just because I couldn't do it without this famous meme... South Park fans will understand..

26fab485e850ff30be5d03a628442dc7ea7821ee7dfbed623ea95e7eb0d3acc1

pietercolpaert commented 10 years ago

In meantime: https://twitter.com/NMBS/status/486067729452269570

pietercolpaert commented 10 years ago

Aaand troubles implementing the API using the new interface as "DIRECTION" is lacking from their new XML interface:

https://twitter.com/iRail/status/486146356957184000

screenshot from 2014-07-07 15 48 19 screenshot from 2014-07-07 15 43 40

ChristopheVersieux commented 10 years ago

Can you do something, even if the direction is missing? You cannot imagine the number of emails I am currently receiving...

Giving at least some results would be really helpfull and avoid the angry crowd of users. The one star review circus for the app has began.

jncn commented 10 years ago

I agree with @ChristopheVersieux, maybe we can put a placeholder as direction? [NMBS Error] or something? Or as the NMBS does: the id, like IC 2237. I pulled Railer from the App Store to avoid one-stars, but my inbox too, is exploding.

pietercolpaert commented 10 years ago

Unknown is now the direction everywhere. Api back online!

jncn commented 10 years ago

Thanks!

ChristopheVersieux commented 10 years ago

Thank a lot, awesome team.

Just for the record, here is just a screenshot of the effects of a few hours without results on the application ratings...

It just break one month of updates, replying user requests, and dozens of hours for the new UI released a few days before the incident.

screen shot 2014-07-09 at 14 57 01

Tuinslak commented 10 years ago

create a nice pdf with the "post mortem" and send it to pr@nmbs.be ? :p

ChristopheVersieux commented 10 years ago

From the feedback I received, there is still the bus station issue as it already happens. Brussels south is the easiest station to test.

The timetable select the bus station as stating point (no platform are displayed)

Removing the app from the store while you fix the issue, I cannot let the app receive so many bad ratings and loose a few hundreds users.

pietercolpaert commented 10 years ago

Hi Christophe,

What exactly do you mean with the bus station issue? I've just tested in the app with "Brussels South" and works perfectly. Maybe this is because I have "English" as my default language?

Pieter

jncn commented 10 years ago

Small idea: can we -- until NMBS adds direction again -- return the last station as the direction, based on this information? http://api.irail.be/vehicle/?id=BE.NMBS.ic52 Or alternatively, built a model with the last stop for every vehicle id and get it from there? I can imagine it would take a little longer before connections are returned, but this is at least better than nothing.

What do you guys think?

ChristopheVersieux commented 10 years ago

@pietercolpaert This simple test: http://api.irail.be/connections.php?to=MONS&from=BRUXELLES%20MIDI&date=100714&time=2200&timeSel=depart&lang=en&typeOfTransport=train&format=json&fast=true

Telles me that IC1743 will leave Brussels south at 1405023960 => Thu, 10 Jul 2014 20:26:00 GMT this API call also give no platform as departure.

While NMBS and iRail (maybe based on ID, not the semantic name "Bruxelles Midi") give the correct result: 22h35 and platform

also, I copied the 2 feedbacks I received because it seems that it also gives some wrong result when using other Brussels station

screenshot_2014-07-10-22-31-38 screenshot_2014-07-10-22-31-54 screenshot_2014-07-10-22-34-05 screenshot_2014-07-10-22-37-03 screenshot_2014-07-10-22-33-49 screenshot_2014-07-10-22-35-15

jncn commented 10 years ago

I've noticed this too, a few weeks ago. Also in BXL midi.

Maybe we should move these API-problems to new and specific issues on the API repository? Only this repo's last commit was back in June 2013...

ChristopheVersieux commented 10 years ago

I can create a new bug entry if necessary. I posted here as it was related to the above fix. I am pretty sure it as working fine before NMBS bug If I understood correctly the current API is a workaround while waiting a fix from NMBS, right? Does this deserve a new bug? In any case the application is currently not ready to be published again publicly. On Jul 11, 2014 12:03 AM, "Jan Cornelissen" notifications@github.com wrote:

I've noticed this too, a few weeks ago. Also in BXL midi. Maybe we should move these API-problems to new and specific issues on the API repository? Only this repo's last commit was back in June 2013...

— Reply to this email directly or view it on GitHub https://github.com/iRail/hyperRail/issues/39#issuecomment-48670633.

pietercolpaert commented 10 years ago

@jncn This is the repository of the API that's going to rule them all. Issues over here are alright!

@ChristopheVersieux I'm going to try to fix it first. If not fixed after today, I'll create a new issue and close this one :)

ChristopheVersieux commented 10 years ago

Hello guys,

I don't want to be rude, but all these issues are causing really problem in my inbox and my phone that rings a lot during the day/evening for the last 10 days (way better since the API temporary fix)..

Should I forward all these emails to info@irail.be ?

@pietercolpaert should I create a new ticket for this specific issues (Mainly Brussels, but I suspect other issues)?

pietercolpaert commented 10 years ago

@ChristopheVersieux No one can explain this madness... but I've seemed to have fixed it. Better fix later.

ChristopheVersieux commented 10 years ago

Ok, Bruxelles is fine with your "hard fix" but Charleroi Sud has the same issue (Charleroi Zuid is fine BTW)

Should I create a separate bug as the resolution already started here?

On Tue, Jul 15, 2014 at 7:47 PM, Pieter Colpaert notifications@github.com wrote:

@ChristopheVersieux https://github.com/ChristopheVersieux No one can explain this madness... but I've seemed to have fixed it. Better fix later.

— Reply to this email directly or view it on GitHub https://github.com/iRail/hyperRail/issues/39#issuecomment-49067741.

pietercolpaert commented 10 years ago

Fixed https://github.com/iRail/iRail/issues/31