[Talk-de] Potlatch discussion (was: Potlatch alpha to the Ubahn U1 in Hamburg)

[Richard Fairhurst]

Hello all,

Please excuse me writing in English - very impolite on a -de list I  
know, but sadly I don't speak German and have had to follow the  
discussion through Google Translation. Sorry.

Could I please reiterate:

- Potlatch will have a history/revert option shortly after API 0.5  
goes live, so that mistakes can be undone. This has already been  
extensively discussed on the dev list if you'd like to read the old  
discussions.

- Potlatch is under very active development and is already becoming  
easier for new users. In recent weeks, a new "practice mode" has been  
added (so you can play without saving changes), and a direct link to  
the documentation. All nodes (POIs) are now also displayed, so that  
users can delete any which are not necessary. There will be a new  
version coming this weekend, as part of API 0.5, with greatly improved  
click handling. In the next few months, the tag-editing UI will be  
redesigned to make it easier for the OSM novice. There will continue  
to be more improvements along these lines.

- Stray nodes/segments and database corruption are generally _not_ the  
fault of Potlatch itself. They are because the server was formerly  
very, very slow. Consequently, when Potlatch issued a request to the  
API, it would often time out and stop halfway through, leaving an  
incomplete write (e.g. nodes written but not ways). Now that TomH has  
implemented quadtiles and the server is faster, this problem is much  
reduced.

- Bad edits are not exclusive to one tool. In my area (Oxford, UK) I  
have to clear up a lot of rubbish left by JOSM users: nodes adjacent  
to ways which have not been made part of the way; long runs of  
untagged nodes with no ways or segments connecting them; segments left  
unwayed for a whole year; etc. Potlatch does have many faults but, in  
fact, all of the above mistakes are more likely in JOSM. Similarly,  
the old Java applet had major bugs that destroyed data (e.g. all tags  
were removed from nodes).

- Potlatch's new automatic simplification (Douglas-Peucker) for GPS  
tracks is also designed to address a very frequent "bad edit" problem  
- that of imprecise, roughly traced roads, especially in rural areas.  
But naturally, because Potlatch is more open to newbies, there are  
likely to be more mistakes made with it. The new revert functionality  
will make it easier to fix these mistakes.

- If you would like the old Java applet to be available again, I and  
others have suggested several times that an interested developer could  
add a user preference ("Potlatch or Java applet") so that users could  
choose the one that suits them. You will now of course have to update  
the applet to understand the 0.5 API but the offer is still open for  
you to do this.

- Some people prefer Potlatch, some people prefer JOSM. Some people  
have machines that cannot run Potlatch (no Flash player), some people  
have machines that cannot run JOSM (no Java 1.5). Having two editors  
makes it possible for users to choose.

- Potlatch is fully open source (public domain) and the code is in  
svn. Everyone is welcome to submit patches and indeed both SteveC and  
TomH have done so in the past. Even if you are not a developer, you  
are more than welcome to make suggestions on the wiki (Potlatch/Bugs),  
using trac, on the dev list, on the OSM diaries, or by mail to me -  
whichever is easiest for you.

- Potlatch has a 't' in it. :)

Please do not think that your concerns are not being listened to, they  
are. I and the other developers want OSM to be both welcoming for new  
users and comfortable for existing users. To do this we have to  
provide a wide variety of easy-to-use tools for people from all  
backgrounds, and the continued work on Potlatch is part of this.

cheers
Richard
Potlatch developer