The JP Datastore files have not been updated this week and it is now Thursday! Bank Holiday blues?
This compounds some issues in last week’s upload, where the gaps included some changes which came in last weekend (42 119 264 for example) and some of the Carnival specials, though I appreciate that the LU strike kiboshed many of them.
The gaps can’t be filled from the Working Timetables either. For example the new 42 Sa and Su WTTs are present but not the MF one. This situation is much more common now than under the old system, though maybe it is because the times at which things are done have changed, rather than because of the change of system itself.
It has now been updated. At first glance it appears to have gained a zip file relating to NR/TfL Rail etc services and lost the one replacing to rail replacement services. But that is at first glance only.
Apologies - there was a delay in publishing to this due to a shortage of team members available as well as the bank holiday. This should return to normal from this week.
What it says on the tin. It’s Wednesday afternoon and still no update of the JP Datastore file.
I did mention this on another thread but that may have been missed.
yes, there’s been a delay due to some processing issues. We should be able to get this revision out soon.
Not on either of the links I use.
Pleased to say it is there now!
probably was the cache that needed to expire!
At least it’s not me that needed to expire. Just to clarify, do you mean a cache at your end? I appreciate that individual pages get cached on my computer and need refreshing but I wouldn’t have thought that applied to a link calling a 130Mb download.
Is there a similar issue with /Line/Route? I’m still seeing a 1/2/2023 date. This is pretty normal for me - nearly everything else seems to get updated days earlier.
I think I have commented before that it would be useful to have an idiot’s diagram showing interdependencies between the various outputs and the main sources they are drawn from. Also when each output is supposed to be updated.
Having said that, my experience of complex systems is that at some point a question gets asked and, rather like on QI, the answer is “nobody knows”!
Hi @mjcarchive @nickp ,
I don’t really have time to do a diagram for you at the current time. We do have internal documentation but not at a high enough level I can share publicly. The data build process is rather monolithic and we’re working on making it more modular so we’re able to do frequent releases and respond to bugs or data corrections.
The Journey Planner Timetables zip (TransXChange) file is normally uploaded weekly on a Tuesday by my colleagues in the JP team. We use a version of this to build our API’s reference data (including Line/Route) and the build/test cycle means this is usually released to production on Wednesday or Thursdays (also weekly).
Sometimes this release can be delayed with technical issues building or getting the build to pass tests. This week, we’re a bit behind due to the file not being available until late Wednesday as well as a fares revision that we had to incorporate.
We are aiming to release this data today and try to get back to our usual timings this week.
Thanks @jamesevans, that is useful background.
If I could ask one more question, does that mean the version used to build the API reference data must always be consistent with what is in the TransXChange file or is there scope for differences to arise?
It should be in general. We don’t consume RRBS schedules into the API and not all engineering timetables.
There are some cases where we’ve had to manipulate files to workaround a bug in processing.
The update is usually done by early Tuesday evening but nothing yet. Bank Holiday week?
Just to say that the file was there by this evening.
Nothing new there yet this week on https://tfl.gov.uk/tfl/syndication/feeds/journey-planner-timetables.zip
That’s just a bit late but my real concern is that the other (preferred) source http://tfl.gov.uk/journey-planner-timetables.zip is producing an error
which makes me wonder whether the update has hit a problem. Full disclosure - I’ve moved fromn Windows 10 to 11 since last week but I doubt that has anything to do with it.
Our linux-pased cron job has failed overnight, but looking at the logs this seems to have been happening for a few weeks.
Think it’s OK now (but the 87 file identified last week remains proudly naff).