Vincent de Phily has commented on the following diary entries

Post When Comment
Do Not Bother to Post a JOSM Bug-Report for a Plugin 19 days ago

It'd be great for somebody to help maintain this plugin.

Some background: Getting the source (no need for an account): svn co Another attempt at a terracer plugin (maybe not the only one):

No need for permission to start hacking the code, only to upload to one of the osm-controled repos. Or if you don't like svn, you could do like derickr and put your version on github or similar. Then work with derickr and pull patches from each other, and don't forget to get your plugin listed in josm once it's in demonstrably better shape than the original one.

BTW I'm not (much of) an OSM dev, I got all this info in a couple of minutes just by searching the wiki and the net, it was yours for the taking as soon as you started getting annoyed that fixing the terracer plugin wasn't happening fast enough.

Do Not Bother to Post a JOSM Bug-Report for a Plugin 20 days ago

My experience of JOSM bug reporting isn't that bad, reporting is certainly worth the time.

Concerning the "use latest snapshot" requirement, it is pretty standard as many projects use this to ward off duplicates and false positives. And in your case it should be pretty easy to grab that josm-latest and report using an up to date stacktrace ?

The problem of unmaintained plugins is frustrating, but it's hard to blame the JOSM devs for code that they didn't write (and apparently don't use) and sits outside their repository. Terracer is really usefull and I'd like to see it in core... But that's a feature request, not a bug report. I'm not sure what can be done to reduce the problem of onmaintained plugins.

Seeing no progress being done on a bug that affects you is of course frustrating. But venting that frustration by re-posting near-identical bug reports (1 2 3 4) is counter-productive. It wastes developer's limited time and hurts motivation. It's not surprising that it feels like spam to them.

If you've got new information, you should add it to the existing report instead of creating a new one, that's a basic principle of bugtracking. If many months have passed with no reply it's acceptable to add a polite "any news ?" comment on a report, but unless you're a paying customer you should be prepared for a "sorry, no time, patches welcome" reply. Sometimes you get no reply, and that has the same implicit meaning.

Let's pretend like contribution is an import. 27 days ago

An import and an editor are completely different things, you can't applie the guidelines of one for the other. Sounds like you're just looking for a justification to effectively ban contribbutors. A simplistic "solution" to a complex problem, hiding it instead of fixing it.

I agree that there are things that do make contributions often sub-par (compared to the average newbie on other editors) : using old / incomplete data, and not interfacing with the osm community as much.

But these differences are not bugs, they are usecases. They need to be handled, not abandoned. Opening up editing to people who can only use offline data, or non-geeks who'd never read a full wiki page is an important thing to do for the OSM project, and I applaud for working on it.

Some issues are being worked on (more frequent data updates, multilingual names...) Some issues are trickyer (Improving the changeset discussion reply rate, foolproofing presets...). Some things are sorely needed but outside the scope of (better tools to detect and help newbies).

Don't condemn, help them. They're just the messenger of things to come, of practices that OSM needs to become ubiquitous. Other apps, like navme, arguably waste more OSM contibutor time than even though they only post notes.

As for the team: great work so far, but a lot more is needed :) And fix bug 3623 ASAP, it'll reduce the community flak.

GPS Coordinate shortener: what3words vs Mapcode 28 days ago

I stay away from w3w, mostly because it isn't free but also because it has only one precision, and is rooted in one language (they make other languages available, but converting is no fun).

IMHO a Lat/Long encoding shoud:

  • be completely free
  • be short enough to remember
  • be language neutral
  • have a variable precision
  • have an error correction

I don't think that having nearby locations share similar codes is that important, since you're going to ask your device to show you the map anyway, and the error correction protects you. But most elligible systems have that property anyway.

The problem is that we have so many competing similar solutions, and that none have a clear popularity advantage. Some are backed by a company, which helps adoption (sadly w3w seems so far to be winning the marketing game) and some are pure free systems.

Another annoyance is that because they are all similar, it's hard to tell one code from another (say an openpostcode one vs an openlocationcode one).

These days I'm kind of tempted to follow openlocationcode. It doesn't define an error correction code, but its Google backing makes it more likely than others to reach critical mass. Country-specific codes like openpostcode are still atractive, though.

Why local assumptions are wrong for an international project 3 months ago

On of my favorite confusing term is "chalet". It's originaly a french word where it indicates a small(ish) wooden house in the mountains. But in English (and OSM) the meaning changed to indicate self-catering individual tourist accomodations.

This kind of problem is inescapable with human languages. All we can do is make sure the wiki explains these well, and that editor's tagging presets are well translated. Ideally they should even be locallised (which is a step above translated), but this requires a lot of work.

Using mobile editor for the first time 3 months ago

Nice writeup. Note that maps updates and app updates are separate things, and that the devs are working towards making map updates much more frequent.

I love that is becoming an editor, it has a lot of potential. But it's not aiming to ever be a full editor. If you need to do anything more that simple POI editing, have a look at vespucci instead.

OSM 3 months ago

Try asking your question on osm's help site or is arcgis's own user forum. I'm not sure I understand your question properly, but it seems more related to arcgis than osm.

I want make island on Nile but Does not appear as required 4 months ago

Some of the issues, in order of importance:

  • The outer way is tagged as waterway=riverbank (equivalent to natural=water water=river). This means that even if the multipolygon is correct and should make the islands visible, there is another single-way polygon around it which doesn't care about the islands.
  • The way you highlighted in your screenshot is not a member of the multipolygon. Add it with role "inner", like the other islands.
  • The polygon is tagged water=river but not natural=water. According to the wiki, water=* is a refinement of natural=water, and should not exist without the later. water=* might work on its own for some consumers (like the default osm rendering), but not all.
  • All islands have apparently been traced from Landsat imagery, but you might want to use Bing or Mapbox, which have a higher resolution.

Happy fixing :)

Making a multilingual map of India using OpenStreetMap data 4 months ago

Great work, good to see OSM's oft-cited strenghs actually applied in practice. And it'll hopefully increase the number of OSM users and contributors.

You say you got the boundaries from "various sources" but point at an opverpass query, so isn't the source just OSM (with some post-treatment to make sure the inda-approved ones are chosen) ?

Mapping private subdivision roads and other gated roads 7 months ago

+1 this calls for improving routing software, not routing/permission data.

Improving the OSM map - why don't we? [12] 7 months ago

iD still can't set anything else than a comment on changesets, so its numerous users can only tag source=* on individual objects (and also tend to overload the changeset comment with stuff that'd be better placed elsewhere, like hashtags).

IMHO taging the objects is useful when things are not straightforward (typically when multiple source:= tags are needed), but that's an optional addition to tagging the changeset.

When you want to know the source of some data, the source*=* tags on an object are never as good as looking up the object's history anyway. Object's source=* tags are rarely updated. The changesets date, user, comment, source, and imagery tags are invaluable and easy to fetch.

Automatic tracing gone wrong 7 months ago

Yup if the lake is big enough, landsat can be used. Not sure how many of those are left to trace ? And yes whatever you autotrace from, a manual check/tweak afterwards is important.

Potlatch editing 7 months ago

An by "other two" I meant level0 and merkaartor of course :p

Potlatch editing 7 months ago

@Richard P2 works fine and it's great to have, it's just that I don't see much advantage to it compared with the other two.

This is more a "tell my why P2 is great, what am I missing" question than a "you should use FooEdit instead" advice.

Potlatch editing 7 months ago


Answered one of you JOSM question on h.o.o :)

Never thought about displaying misc GPX attributes, but now that you mention it, it would be usefull and I didn't find a way to do it (outside of preprocessing the file. Same behavior in iD (actually even less details displayed), and the only way I managed to open a gpx file in P2 is using the task button but it's arguably even less usefull ?

To each his own for viewing object tags and relations, I very much dislike iD's UI for this, and feel that P2's advanced (aka usable :p) view is pretty much the same as JOSM's.

I cetainly don't want to press people into using this or that editor, I was just curious about P2 because I thought that JOSM and iD and Vespucci already catered nicely to most usecases.

Potlatch editing 7 months ago

Out of curiosity: why Potlatch nowadays ? Just because you're used to it ?

Profile 7 months ago

The best way to fix this is editing the map yourself. Go to the location, draw the buildings (at least your one, but go ahead and draw the whole street :) ), add and address (housenumber and street), and final add the business type and name.

If you gave it a try and still feel it's too daunting, add a Note at your busines's location and wait for somebody to do the edit.

Should we teach JOSM to first-time mapathon attendees? 8 months ago

There's probably a correlation between what makes newbies start with JOSM instead of's default and what increases contributor retention time. The newbie being more technology-savvy and the mentor being more enthusiastic with JOSM spring to mind, but these are just guesses.

Maybe we should assign first-time editors at random during some mapathons, and ensure that fans of each editor are available for mentoring ?

The most inefficient way in North America 8 months ago

I was surprised to not find any overnodeing checks in the main QA tools (OSMI, Osmose, Keepright...) Maybe you could try to integrate your code in one of these, to reach a wider audience ?

OpenStreetMap Foundation Chairperson's Report for the Annual General Meeting 8 months ago

The waiver is great, but it can be a daunting procedure, elegible people might think they aren't, or be too proud to think they are, etc.

Maybe a subcription fee indexed on the country's GDP per capita (or some similar metric) would make sense ? Take the UK as the reference fixed in £, and adjust other countries's fees during yearly AGMs ?