[OSM-talk] Potlatch :-)

[Ulf Lamping]

Richard Fairhurst schrieb:
> I know what you mean, but it's not actually a simple comparison.
>   
Well, compare it with other document centric applications (office or 
alike), you get the same point.
> In some ways the current Potlatch model is pretty similar to  
> Mediawiki (or Usemod Wiki, or whatever), despite the lack of an  
> explicit "Save" button. The "atom" is a page in Mediawiki, or a way  
> in OSM. It is a single discreet unit with minimal dependencies and  
> which can largely be viewed in one edit window. In half an hour's  
> editing, you might edit 30 Mediawiki pages or 30 OSM ways, and  
> there's an explicit save action for each one - albeit that Potlatch  
> implements it through deselect, not a button.
>   
That's one possible point of view. I could also compare the OSM ways 
with words/sentences in the MediaWiki page, so the "Save button model" 
can easily be applied.
> There is absolutely no way I'm going to require users to click "Save"  
> per way/POI - that would be a usability disaster. 
ACK
> So any such change  
> would have to be a "commit unsaved changes" model. But that has  
> serious UI issues too. 
Nothing comes for free ;-)
> Your changes might have been made half an hour  
> ago and be 25 miles off-screen, so you don't actually know what  
> you're saving, whereas now it's all visible on-screen. 
The same applies to very long wiki pages.
> You have to  
> have conflict management: that's really hard for a newbie to get  
> their head around, never mind the 412 errors, and runs the risk of  
> introducing severe db problems if there are coding mistakes or  
> oversights. And so on.
>   
Come on. I'm doing OSM for more than half a year and had exactly one 
conflict problem, you can have sever db problems with one way done 
wrong, ...
> I do see where you're coming from and it's something that I've been  
> considering since Potlatch was first in development - some of those  
> at the Oxford dev day when we finished the Rails port will recall the  
> discussions. But it's not a simple decision.
>   
Of course not.
> To my mind the recent changes to Potlatch (removal of anonymous  
> edits, introduction of undelete/history, fixing "revert merged ways"  
> issue) go most of the way. There is still the issue that you are not  
> explicitly forced to choose between practice and edit mode (i.e. it  
> defaults to edit mode), with some strong on-screen documentation  
> exhorting you to practise first; this is on the list, I just haven't  
> got around to it yet. (And more documentation is needed in general,  
> but that's a whole-OSM problem.)
>   
IMHO that's only a workaround for a clean "Save button model". However, 
having an undo mechanism is a good thing regardless of the save model.
> I guess the reason I suddenly stopped being "mild-mannered" on the  
> subject (sorry Frederik :) ) is that the continual sniping from a few  
> people, historically either on talk-de or from German users on talk,  
> bears absolutely no relation to the reality that I see on the map -  
> and I do a lot of editing across a big swathe of England and Wales.  
> There really is _no_ systemic problem with Potlatch edits. As I've  
> said before, most of the bad edits I see (low-resolution ways, stray  
> nodes) actually originate from JOSM users, and I'm not blaming the  
> tool for that.
>   
I can understand you for being annoyed by such repeating mails. But I 
can also understand mappers getting angry when they noticed things are 
wrecked up/deleted and all that you see is that word beginning with p.

The things about JOSM you've mentioned aren't nice but doesn't really 
hurt and can easily be recovered. If someone deletes/moves ways the 
current tools to recover that are virtually none existing for the 
ordinary user - so those changes are much more painful to recover.


All in all, I'm still convinced that having a "Save" button would be 
much more what the users are used to and therefore expects here. And 
that's my key concern, the user expects an explicit save action (at 
least I do), so he doesn't expect that any changes immediately go online.

Regards, ULFL