I guess it’s time for another short report from the engine room:
With upcoming Core Update 153, where no testing version is available yet, we plan to finally introduce the more precise location database, but this time without breaking systems as we did before.
Unfortunately, this requires a different output for the
xt_geoip module, as it cannot handle tree, but only lists of networks, which must be flattened so there are no overlapping entries (
xt_geoip does not look for the best match, but for the first one). @ms has worked several days on this to make it work fast enough even on dated hardware, and we believe we are somewhat there by now.
While Core Update 153 will contain the fix for the timezone problem (bug #12514), we have had to stop the generation scripts for the current databases, which is why every user now gets a 410 - gone HTTP error.
This is intended and unfortunately the only possibility we have. If we simply publish the more precise database, a number of systems running Core Update 152 or below would break again, and since all
location update clients query the same URL, we cannot distinguish them at this level.
Worse, we cannot generate the more accurate and the less accurate version at this time, since the location clients query those by a DNS record, and if the generation time of the downloaded file do not match that record exactly, the files won’t be processed. It seems as we have built in so many safeguards into the updating procedure that we somewhat crippled ourselves when it comes to this problem.
To keep it short: A testing version of Core Update 153 will be available soon (personal ETA: next 14 days) and as soon this will be released, any IPFire machine running it will get the more precise location database again, regardless of the time zone it is located in.
Sorry for all this confusion.
Thanks, and best regards,