Dear users,
Version 0.7.62 of the Overpass API has been published already a month ago, but I have not announced it yet. This is because the documentation lagged.
I’ve now completed the documentation to also cover
user_touched constraint in a new section about user constraints in generalCheers,
Roland
Has there been any thought given to making overpass work like other Linux packages? Where I could install with a package manager, edit some configuration files, and run a system service? Where log files are placed in a standard location?
I’ve written one of the guides for installing overpass and I still find it exquisitely difficult to install, administer, and troubleshoot.
If it were easier to install, run, and administer overpass, it would take considerable burden off the public instances because people would more easily be able to run their own instances on their own local hardware.
Yes.
There are quite a number of platforms I might publish to. There is no clear priority, and each of them is nontrivial work.
Are you aware of my Docker repo? This has been quite a lot of work: understanding the paradigm of Docker (one main process per container, Docker volumes, user role and rights concept etc.), writing instructions as well as building and running a testing environment. The reason I’ve chosen that distribution channel is that numerous support requests came in about a poorly designed Docker container by a third party.
In the end it had been three month worth work of volunteer time, and there is no clear indication of anyone using that. Opposed to many actual features with more users on less effort to implement them.
I have considered to publish for the Debian and Ubuntu path, but faced a number of open questions. For example the best practice to bring a new version out of band to the users and how to get security updates through. If I compare that to JOSM where the Debian channel is also a niche, I would expect another place of relatively much effort for relatively little benefit.
Other than that, I very grateful for your additional documentation. I’ve got quite some insight how people might attempt to install and run the system, and one result has been e.g. that one can now stop everything safely by Ctrl+C aka SIGTERM.
I think the actual packaging and publishing is not really that important. I see the challenge as more about making the software itself easier to configure and administer and standard ways that *nix administrators are used to.
Thanks for this! I was not aware of it and it sounds from first glance like a major improvement for administrators and may even solve many of the challenges I’m facing.
Oh, that’s the one I thought Roland was talking about - @Kai_Johnson linked it to me in the middle of posting here and silly me for not clicking the link. Yes – that’s what’s needed, a properly configurable overpass where you can just give it the settings and not mess with the all the weird intermediate processes. Looking forward to trying it!