Hi community!
I’m having an issue with another (otherwise excellent) mapper in Exeter who seems convinced all addr tags on building ways in the city should be removed, and converted to new nodes. Their changes are clearly scripted, and offer no meaningful improvement in data.
Unfortunately, after prior conversation we seem to be stuck in a revert war.
Example changeset: Changeset: 186591672 | OpenStreetMap
I need help here - thanks!
If they’re not responding to changeset comments I’d report to the DWG.
Other changesets suggest this was already raised with DWG about six months ago, and the issue has been ongoing since at least this changeset (which includes some discussion on the matter) Changeset: 174204545 | OpenStreetMap
These changeset comments usually cite these changes being made in accordance with “local style", but it’s not clear who has agreed to that or what areas it apparently applies to. In my opinion it doesn’t make sense to have a different standard for doing addresses in one particular city for no apparent benefit. Residential houses with a single address should carry their housenumber and all associated details on the building way, as is done in the vast majority of the UK. The other, less common address style is tagging housenumbers only on entrance nodes, this is neither of those. Tagging @JassKurn for visibility.
To me it makes sense to have businesses and POIs as separate nodes on buildings, but not residential house addresses.
Does 3 people’s style really make a local style? How local is local?
The edits are being made by me, thanks for notifying me @LordGarySugar.
Below is an attempt at a short answer, but it isn’t.
First off…
As with most things in OSM there is no agreed way to tag addresses in the UK, just discussions that fail to reach a conclusion. eg the recent discussion about addr:place
Whether it is best to add addresses to nodes or building ways or doors, is for a separate discussion. Discussing that here will mislead, and make this discussion more difficult to follow
Please start another thread to discuss UK address tagging - Nodes vs Buildings vs Doors.
“A short version”
This is an Exeter issue. The long time main, on ground mappers, in Exeter, map addresses as nodes. (Guy, Wilpin & myself).
There are lots of gaps along streets due to numerous buildings not having house numbers visible. Most of the address mapping is done by Guy & Wilpin, I add a fair few, but mostly concentrate on rural ways.
A few of years ago a likely armchair mapper appeared in Exeter for a short while, and moved vast amounts of the historic node addresses to buildings, likely using a josm extension, then added missing addresses by assuming ((interpolating) addresses, likely using a josm extension. Adding addresses that were not visible on the ground is why I make assume armchair mapping.
I spotted several addresses I had added had been moved to buildings, and addresses created for buildings that I knew had no address visible from the road. This mapper did not see an issue with moving addresses or assuming (interpolating) addresses. After a while this mapper moved on..
These “poor” edits should be moved back to nodes, but I keep leaving it because I felt it needed an on ground survey for the assumed (interpolated) addresses, and I don’t enjoy walking up and down urban streets. I did a bit of mapping and found that although josm had ways to move an address from a node to a building, it didn’t have a way to move an address from a building to a node. Due to extra work needed, I kept putting it off.
Then last year rskedgell announced a project to add UPRN / Postcode data. The plan was to add it to buildings.If you look at that discussion you will see I agree with the import, but not the default position of adding the data to buildings, especially where the local preference was for adding addresses as nodes. He agreed to avoid adding UPRN to buildings in areas where he could see addresses were on nodes, and agreed his imports could be moved to nodes (I’m assuming this should be by local mappers)
This resulted in me realising I had to quickly sort out the area in Exeter where the hit-and-run mapper had moved addresses from nodes to buildings, and guessed (interpolated) other addresses. I sorted out a way to do this in josm and started editing. Unfortunately my first changes happened a few days after Joe-w had done a very small amount of address mapping in East Exeter which involved adding some addresses to buildings, and moving some addresses from nodes to buildings. I believe Joe-w believed my edits were aimed at reverting his edits without discussion, when it was simply coincidence. He reverted my edits without discussing this with me. I responded by saying he should not have done that, he should have contacted me first.
The result was a few reverts. I contacted the Data Working group twice, but had no response. I assume they didn’t want to get involved in a local dispute. I had to stop the mapping and reverts due a few different reasons. But did plan to get back to the issue later this summer..
Then rskedgell started adding UPRN data to buildings in Exeter last week. So I’m started again.
To summarise.
Thanks all. I’ve reverted the changes on the basis they’re undiscussed, automated, and not in favour anyway (based upon the wiki and opinions in this thread).
@JassKurn Please can you discuss your changes before writing scripts and going ahead. Thanks!
(Previous reply to wrong post)
Where was the mention that the changes from @JassKunn were automated? I seem to have missed it.
I also occasionally map in Exeter. While I found the local addressing approach a little idiosyncratic, I do appreciate the consistency of any building being just a building and that an address point is separate. Whether for 1 or many. I kept reading the various guidance and came to the conclusion that there was no conclusively defined way to map an address. Just different conventions. And as I said here recently, we do odd in Devon.
For the record, I’m absolutely fine with both address styles! Any improvement to OSM is great.
I’ve only taken an issue here as ~8,000 ways were batch converted in the space of an evening, without meaningful improvement (like adding data) or any prior discussion before doing so.
There’s a wiki page which advises against exactly this, see here.
To me, it’s clear this was done via automation, which is also discouraged without prior discussion. In this changeset from a few months back, they mention using the JOSM scripting plugin:
Thank you.
So why not go back further to undo the automated mapping carried out by the flybynight mapper @JassKurn mentioned? Especially if the actual address method is not the concern. You could have been supportive of established local mappers instead.
So for the rest of us… could you clarify the ‘flybynight’ mapper, and the ‘established local mappers’? Thanks…!
I’ve reverted your edit.
I’ve done this on the grounds that.
You quote a wiki page and argue that it means we should not change, but ignore you’d had moved addresses to buildings a year ago, just before I started. If moving addresses is an issue why did YOU do it? and without asking? Explain why you move these addresses to buildings?
As I said a mapper moved a vast amount of addresses from nodes to buildings a few years ago, without discussion, and using what you’d call “a script” and appears to have guessed the missing addresses. Why have you ignored that? Is it because you’ve done the same thing? (see your edits above)
And the scripting argument is a nonsense. You just used a revert script without discussion? As I said before josm (and Id?) has tools (scripts) to simplify moving addresses from nodes to buildings, but not from buildings to nodes. I simply reversed the available extension principle.
I think my editing history in OSM will show that I’m bit of a pendant, and having about 20 years of history here I am well aware of how easy automation leads to problems. My method was very cautious.
I agreed that my very first edit was too big and should have been broken up into smaller areas, so I accepted your very first revert, and since then have been doing them in relatively small areas.
Regarding my own mapping - yes, sometimes I delete address nodes! Usually because there’s only a few sporadically placed (sometimes inaccurately), and I’m replacing them with a full street of aligned buildings with address data. Often after a survey. I hope you’ll agree these are worthwhile additions. 
This is what I mean when I ask for no conversions, unless done so alongside a clear improvement.
Sorry, the phrase used above was ‘hit and run’. Can’t clarify more as I’m not investigating the changesets.
Established. I’m a new mapper. I look to those who have mapped locally for longer and well before me for help and advice. That’s how community develops. The established mentor the less established. Catch the errors etc.
@JassKurn I’ve reverted this again because you seem to be converting large parts of the city.
If there’s buy-in from others here, could you at least:
When I’m mapping in Exeter, I’m happy to stop removing address nodes, as I can understand the hypocrisy in doing so.
Of course, it’s near impossible to define the term ‘local’ of ‘local style’, so this feels like a fair outcome. What do you think?
Just curious, what it one day all the address nodes can be imported from the government database. (Surely there is one.)
That’s why (in Taiwan) I’ve not bothered to add anymore addresses by hand.
That’s not going to happen in the UK, as that would impact the ability of a foreign-owned monopoly to extort money.