| Commit message (Collapse) | Author | Age | Files | Lines |
|---|
| |
|
|
|
|
|
|
|
|
|
|
| |
this quirk was forgotten during the haskell rewrite (reported by
networkException).
It's now done in a somewhat awkward way: it's hard to do this before
parsing the csv (when it's still an unencoded ByteString), but doing it
later might easily miss relevant places. In either case this does lots
of slightly unecessary copying 🤷
Also the nested quadruple fmap feels like its own justification
|
| |
|
|
|
|
|
|
|
|
|
|
|
| |
well this blew the query up ..
it attempts to do two things:
- find platform sections grouped under platform_edges (important if
platform sections of adjacent tracks don't line up)
- if that fails, find any platforms somewhere under the whole platform
object
This should (hopefully) cover most cases where platform sections are
actually mapped at all.
|
| |
|
|
| |
esp. a trailing / as in bahnhof.name/MH/ should not lead to ??.
|
| |
|
|
| |
this should now work with NixOS 23.11's hackage snapshot
|
| | |
|
| | |
|
| | |
|
| |
|
|
| |
seems to give slightly better results
|
| |
|
|
| |
whoops, forgot this bit
|
| | |
|
| |
|
