On Sun, 1 Oct 2023 at 23:16, Yuchen Pei <i...@ypei.org> wrote: > > The preparation and planning is well progressed in my view. There is > > always going to be a long tail of corner cases and I was attempting to > > handle more of these in the code and that never got finished. We > > probably would be better to make a call to accept those corner cases > > and go ahead with the import. > > Thanks for the context. Which of the 7 Stages mentioned in the README > correspond to the corner cases? >
It's been a while since I worked on this, but I believe it was the matching of existing OSM addresses to Vicmap, and that matching affects most of the import stages. > > Are you interested in working on it? > > Absolutely! Thankfully git is decentralised so I don't really need a > gitlab account. It would be much easier if you can create one though as otherwise collaboration will be difficult. From what I can see you just need either a phone number OR credit card, so you should be able to create an account without a credit card. On Mon, 2 Oct 2023 at 11:45, Yuchen Pei <i...@ypei.org> wrote: > Maybe we can go stage by stage. Stage 1 - postal_code is the first stage > that has not been completed. > > I built the dist/postalCodeURLs.txt file (see attached) yesterday and it > contains 2425 JOSM RemoteControl urls. Shall we go ahead and import > them (or a newly generated version)? > I spent some time today reviewing the import again, and this one was ready, so I've done it at https://www.openstreetmap.org/changeset/142031616 For context, we decided not to include addr:postcode on each address object imported, therefore to ensure we can still capture address postcodes, they are added to the admin_level=9 boundary relations. There were a bunch already mapped but this changeset completed those missing the tag. My prior analysis comparing these boundaries in OSM and the Vicmap address points with postcode found this is a reliable way to ensure we have postcode coverage, except for some edge cases mentioned in the import documentation. > BTW, I'd assume the whole pipeline for each stage should be generated at > approximately the same time, using the data downloaded at approximately > the same time. If that is the case, it would make sense to have a > Makefile rule for each stage, that remove all files in the pipeline for > that stage and redownload and recompute them. What do you think? > It was built to run the whole process and generate all the outputs from all the import stages together, at this stage I wouldn't be refactoring it.
_______________________________________________ Talk-au mailing list Talk-au@openstreetmap.org https://lists.openstreetmap.org/listinfo/talk-au
