rth
commented
4 years ago Thanks for opening this issue. The issue with http://download.geonames.org/export/dump is that last time I checked it didn't actually have files with exactly the same format as https://download.geonames.org/export/zip/ so we can't use it as a cache.
I think a better solution would be to store these files in a separate github repo, and then serve them via Github Pages as an alternate location. This would have the added advantage,
- that we can do proper versioning of postal code date between different geodes updates (to see what changed)
- potentially add patch files with missing or incorrect postal codes in the same repo (.e.g. https://github.com/symerio/pgeocode/issues/23)
I have actually started https://github.com/symerio/postal-codes-data last time this happened but I think it may have files from the wrong URL. Also GB_full is too large to be added on Github unfortunately.
So what we would need is,
- a Github action on https://github.com/symerio/postal-codes-data to periodically download data from geonames.org and put it into this repo
- a mechanism in pgeocode to specify alternate URL locations.
Would anyone be interested in looking into it?
dshinzie
With issues #40 and #34 , we see that genonames data source that we rely on seems to have intermittent availability issues. To protect against such intermittent failures can we think of adding a fallback data source or a cache. Both the issues were short lived and resolved by itself.
In Issue #40 @richunger pointed out that the data was available at http://download.geonames.org/export/dump/ when we had the issue, can we rely on that as a backup data source.