GNOME Bugzilla – Bug 723463
Fix typo in geocode-forward place_type_from_attributes
Last modified: 2014-02-04 11:00:40 UTC
The class name is boundary not boundry. Patch forthcoming.
Created attachment 267832 [details] [review] forward: fix typo in place type The class name is boundary not boundry.
Ok, so this caused segfault in Maps. The code inside the if-statement have obviously never been run. The check in the hash_table for place_rank always returns NULL. And this has an explanation: place_rank is not returned in the json of &format=json. It is however in &format=jsonv2. So to get place_rank we need to switch to jsonv2. Another difference I spotted between json and jsonv2 is that "class" is called "category" in v2. So maybe a patch that switches to jsonv2 is what is needed.
Created attachment 267836 [details] [review] forward: switch to jsonv2 format for Nominatim The json format does not include the place_rank attribute. The place_rank attribute is needed to establish what kind of boundary we are dealing with. Switching to jsonv2 gives us the place_rank at the cost of having to rename the class attribute as category.
Created attachment 267838 [details] [review] forward: switch to jsonv2 to include place_type The json format does not include the place_rank attribute. The place_rank attribute is needed to establish what kind of boundary we are dealing with. Switching to jsonv2 gives us the place_rank at the cost of having to rename the class attribute as category. This patch also fix a typo where boundary was spelled as boundry. Ensuring that the missing place_rank attribute never was detected. If fixed in a stand alone patch it exposes us to a commit that introduces a segfault.
Review of attachment 267838 [details] [review]: ACK
Attachment 267838 [details] pushed as 49ba62c - forward: switch to jsonv2 to include place_type