Re: [OSM-talk] Undo request button for changesets

2009-07-20 Thread MP
> A brilliant idea. Close inspection of a changeset should also be able to
>  better find out whether the changeset actually applies to a certain area
>  or whether it just has a huge bounding box.

Currently, we have "ID, Saved at, Comment, Area" fields in the
changeset list, so perhaps adding columns with number of ways, nodes
and relations changed/added/deleted would make it easier to guess the
nature of a changeset. Unusually high number of moved nodes (in
relation to other edits in changeset suggest thatr someone may have
moved something, perhaps accidentally). Unusually high number of
deletion suggest possible vandalism (though it can be case that
something big got just bolldozed in reality, but it is worth
checking), etc ..

Martin

___
talk mailing list
talk@openstreetmap.org
http://lists.openstreetmap.org/listinfo/talk


Re: [OSM-talk] Undo request button for changesets

2009-07-20 Thread Frederik Ramm
Hi,

Martin Koppenhoefer wrote:
>> On 16 Jul 2009, at 13:48, Sam Vekemans wrote:
>> Umm..  Not everyone is that friendly and we need more powerful medicine for
>> those occasions.
> 
> +1
> 
>> 1) A way to view quickly determine the nature of any change-set - did the
>> change set add any new ways/nodes? did it add tags to ways/nodes? change
>> values in tags?  delete ways/node? delete tags? - that sort of stuff.  I
>> would like to see all the changes textually and also as a map and be able to
>> see the history of each feature with similar information available.
> 
> +1
> yes, it would be really helpful to get in the object history a map not
> just of the current state but on demand of the other (previous) ones
> as well (could maybe be rendered locally?). By this you would have an
> easy and in many times sufficient way to judge the nature of an edit.

A brilliant idea. Close inspection of a changeset should also be able to 
better find out whether the changeset actually applies to a certain area 
or whether it just has a huge bounding box.

This is by the way, just to re-iterate my mantra, not something that for 
whatever strategical reason needs to be done on the main server(s). It 
can be done by anyone running a current mirror. We could even have 
multiple tools like these, which could shine a light on different 
aspects of editing and employ (and try out) various algorithms of 
highlighting changes.

Bye
Frederik

___
talk mailing list
talk@openstreetmap.org
http://lists.openstreetmap.org/listinfo/talk


Re: [OSM-talk] Undo request button for changesets

2009-07-20 Thread Martin Koppenhoefer
2009/7/20 Peter Miller :
> On 16 Jul 2009, at 13:48, Sam Vekemans wrote:
> Umm..  Not everyone is that friendly and we need more powerful medicine for
> those occasions.

+1

> 1) A way to view quickly determine the nature of any change-set - did the
> change set add any new ways/nodes? did it add tags to ways/nodes? change
> values in tags?  delete ways/node? delete tags? - that sort of stuff.  I
> would like to see all the changes textually and also as a map and be able to
> see the history of each feature with similar information available.

+1
yes, it would be really helpful to get in the object history a map not
just of the current state but on demand of the other (previous) ones
as well (could maybe be rendered locally?). By this you would have an
easy and in many times sufficient way to judge the nature of an edit.

cheers,
Martin

___
talk mailing list
talk@openstreetmap.org
http://lists.openstreetmap.org/listinfo/talk


Re: [OSM-talk] Undo request button for changesets

2009-07-19 Thread Peter Miller


On 16 Jul 2009, at 13:48, Sam Vekemans wrote:


Hi all,
+1 for having a revert button only for your own (personal) changsets.
With no button requesting a change revert. .. just message that  
person, in a friendly way.

Most of the time it's a newbie who is just learning. :-)


Umm..  Not everyone is that friendly and we need more powerful  
medicine for those occasions.


We have a persistent and very destructive mapper in the UK. We have  
reverted most of his edits from June by hand. I messaged him politely  
in June without a response. He reappeared in July and is now breaking  
roads, rivers and railways and adding stray bits of visible rubbish  
most days. All  of this stuff is majorly destructive:-

http://www.openstreetmap.org/user/liam123/edits


I think we need a number of things pretty fast.

1) A way to view quickly determine the nature of any change-set - did  
the change set add any new ways/nodes? did it add tags to ways/nodes?  
change values in tags?  delete ways/node? delete tags? - that sort of  
stuff.  I would like to see all the changes textually and also as a  
map and be able to see the history of each feature with similar  
information available.
2) A way to for people to patrol for edits by new contributors in an  
area on a real-time basis and a way for experienced mappers to  
indicate that a newbie's changesets has been checked and are ok or not.
3) An RSS feed to monitor a users edits to watch for new activity. Got  
it, we have that one - thanks
4) A way to revert a change-sets quickly before people put stuff on  
top or try to repair (I would also like to opportunities to complete  
the revert even if work has subsequently been done on some of the  
touched features)
5) A way to see which change-sets have been reverted recently within  
an area and by whom

6) A way to revert the revert etc
7) A way for people spot revert wars (ie users doing many reverts or  
areas with many reverts)
8) Community guidelines on only using revert for blatant vandalism  
rather than just because you preferred it as is was - (assume food  
intentions and build if in doubt)
9) A way to request a block on the IP address for a short period if  
vandalism is coming from  unique static IP address
10) I would also like a publicly available talk page for a user where  
people can leave public messages for the user and for other people to  
read. This would need to be a wiki page so problems could be removed,  
but it would allow a place to talk in semi-public about issues which  
is sometimes useful.


I think all of the above is the what we need to have to make it work  
from a community perspective as well as from a technical perspective.


Do be aware that a vandal is very unlikely to sign the new license so  
at present it is important that we remove the edits rather than repair  
on top of edits.





Regards,



Peter Miller




(If we had a "revert" button next to a changeset, and I agree with the
cautious voices who said that this is quite unlikely to work once a  
few
hours or days have passed, then I would probably enable that button  
only

for the user who committed the original change, and anyone else who
clicks the button would just send a message to that user saying  
"please

revert your change".)

Bye
Frederik

From my experience, the user who did the upload. .. ie, if i  
uploaded something that i realize that more nodes were added than  
needed.  Having an easy way for me to fix that is helpful.
I found that once i notified users of what was done, they  
(themselves) would probably appreciate the ability for them to  
revert their mistake (since now becoming aware of it).   ie. in the  
case of copyright maps.


For example, im now fixing up some of the stuff i added for my  
sample area.  My changesets are in a logical order.  So i'd like the  
ability to be able to 'undo', something i did last week.
So if other edits where made ontop of some of the changes, a list  
can be made available of those conflicts. But all the other stuff  
(breadcrumbs) can be reverted.  (some of my polygons might have  
extra nodes, that i dont know of yet)


Cheers,
Sam
___
talk mailing list
talk@openstreetmap.org
http://lists.openstreetmap.org/listinfo/talk


___
talk mailing list
talk@openstreetmap.org
http://lists.openstreetmap.org/listinfo/talk


Re: [OSM-talk] Undo request button for changesets

2009-07-16 Thread Sam Vekemans
Hi all,
+1 for having a revert button only for your own (personal) changsets.
With no button requesting a change revert. .. just message that person, in a
friendly way.
Most of the time it's a newbie who is just learning. :-)

(If we had a "revert" button next to a changeset, and I agree with the
> cautious voices who said that this is quite unlikely to work once a few
> hours or days have passed, then I would probably enable that button only
> for the user who committed the original change, and anyone else who
> clicks the button would just send a message to that user saying "please
> revert your change".)
>
> Bye
> Frederik


>From my experience, the user who did the upload. .. ie, if i uploaded
something that i realize that more nodes were added than needed.  Having an
easy way for me to fix that is helpful.
I found that once i notified users of what was done, they (themselves) would
probably appreciate the ability for them to revert their mistake (since now
becoming aware of it).   ie. in the case of copyright maps.

For example, im now fixing up some of the stuff i added for my sample area.
My changesets are in a logical order.  So i'd like the ability to be able to
'undo', something i did last week.
So if other edits where made ontop of some of the changes, a list can be
made available of those conflicts. But all the other stuff (breadcrumbs) can
be reverted.  (some of my polygons might have extra nodes, that i dont know
of yet)

Cheers,
Sam
___
talk mailing list
talk@openstreetmap.org
http://lists.openstreetmap.org/listinfo/talk


Re: [OSM-talk] Undo request button for changesets

2009-07-16 Thread Lambert Carsten
On Wednesday 15 July 2009 16:36:36 Russ Nelson wrote:
> On Jul 15, 2009, at 7:10 AM, Frederik Ramm wrote:
> > Hi,
> >
> > Yann Coupin wrote:
> >> I completely agree with Pieren here. Unless you're part of the happy
> >> fews, if case of vandalism/error, you're forced to painstakingly
> >> repair by hand the problems
> >
> > That is true, but we also have to look at the potential for problems.
> >
> > ("If everyone had a gun, it would be much easier for everyone to
> > defend
> > themselves, and the streets would thus become much safer for
> > everyone."
> > - Can you spot the problem?)
>
> No, in fact, I can't, because you're using the wrong analogy. Try this
> one:  "If only police and criminals have guns, it is much easier for
> everyone to defend themselves, because when a criminal is threatening
> you with his gun, you just call a policeman, wait for him to arrive,
> and he uses his gun."
The gun analogy is a very bad one whichever way you use it. In the real world 
there is no undo/revert. Philosophical nitpicking aside we can undo/revert 
things in the data so there is no real risk involved unlike the physical 
world. Having very powerful tools on either side will do more good than bad.
>
> > If we make these things too easy, then they will get abused.
>
> Do you realize that you're also making this argument:
> "If we make map editing too easy, then abusers will edit the map, thus
> we must put horrible user interfaces in place to make it harder to
> edit."  One might argue that this has already happened.  :)  (just
> teasing!)
>
> > Also, it is quite a challenge technically as well. A good revert
> > system
> > would have to visualise what it is about to revert,
>
> Fair enough!  Perhaps a revert should be couched in terms of a .OSM
> file which you load into JOSM, and then preview?  If you like what you
> see fixed, then you hit the upload button.  If there are conflicts,
> well, then there are conflicts and you deal with them as any other
> edit conflict.
Agreed.
Ordinary mappers like me need an easy way to download the deleted objects so 
they show up in Josm (or other).

Lambert Carsten


___
talk mailing list
talk@openstreetmap.org
http://lists.openstreetmap.org/listinfo/talk


Re: [OSM-talk] Undo request button for changesets

2009-07-15 Thread Russ Nelson

On Jul 15, 2009, at 7:10 AM, Frederik Ramm wrote:

> Hi,
>
> Yann Coupin wrote:
>> I completely agree with Pieren here. Unless you're part of the happy
>> fews, if case of vandalism/error, you're forced to painstakingly
>> repair by hand the problems
>
> That is true, but we also have to look at the potential for problems.
>
> ("If everyone had a gun, it would be much easier for everyone to  
> defend
> themselves, and the streets would thus become much safer for  
> everyone."
> - Can you spot the problem?)

No, in fact, I can't, because you're using the wrong analogy. Try this  
one:  "If only police and criminals have guns, it is much easier for  
everyone to defend themselves, because when a criminal is threatening  
you with his gun, you just call a policeman, wait for him to arrive,  
and he uses his gun."

> If we make these things too easy, then they will get abused.

Do you realize that you're also making this argument:
"If we make map editing too easy, then abusers will edit the map, thus  
we must put horrible user interfaces in place to make it harder to  
edit."  One might argue that this has already happened.  :)  (just  
teasing!)

> Also, it is quite a challenge technically as well. A good revert  
> system
> would have to visualise what it is about to revert,

Fair enough!  Perhaps a revert should be couched in terms of a .OSM  
file which you load into JOSM, and then preview?  If you like what you  
see fixed, then you hit the upload button.  If there are conflicts,  
well, then there are conflicts and you deal with them as any other  
edit conflict.

--
Russ Nelson - http://community.cloudmade.com/blog - 
http://wiki.openstreetmap.org/wiki/User:RussNelson
r...@cloudmade.com - Twitter: Russ_OSM - 
http://openstreetmap.org/user/RussNelson


___
talk mailing list
talk@openstreetmap.org
http://lists.openstreetmap.org/listinfo/talk


Re: [OSM-talk] Undo request button for changesets

2009-07-15 Thread Simon Ward
On Wed, Jul 15, 2009 at 03:37:35PM +0200, Frederik Ramm wrote:
> Pieren wrote:
> > It is a technical discussion because everybody can revert changes anyway.
> 
> True. Everyone can delete all of London anyway. This doesn't mean that 
> we have a button on the web page that says: "Delete all of London".

That’s quite an exaggeration over reverting the edits made in a changeset,
committing in a new changeset, which itself can be reverted.

If somebody did delete all of London using the mechanism above, I would
welcome the option to revert that all the more.

> (If we had a "revert" button next to a changeset […] then I would
> probably enable that button only for the user who committed the
> original change, and anyone else who clicks the button would just send
> a message to that user saying "please revert your change".)

That breaks down if the user fails to respond, and it’s back to the
admins to evaluate the situation and revert, or potentially spending a
lot of effort manually reverting the change.  For a crowd‐sourcing
project that thrives on a low(ish) barrier to contributing, making tasks
unnecessarily difficult seems to be going entirely in the wrong
direction.

Of course, it’s polite to let someone know if you’d like their change to
be reverted, but you can still do that if a revert button exists next to
a changeset.  There are links to the user page at the moment, but
messaging can be encouraged by providing a direct “message this user”
link too.  Leaving the option of reverting a changeset (if possible
without conflicts) to the users is much more open than leaving it to a
privileged few.

Simon
-- 
A complex system that works is invariably found to have evolved from a
simple system that works.—John Gall


signature.asc
Description: Digital signature
___
talk mailing list
talk@openstreetmap.org
http://lists.openstreetmap.org/listinfo/talk


Re: [OSM-talk] Undo request button for changesets

2009-07-15 Thread Russ Nelson

On Jul 15, 2009, at 6:51 AM, Pieren wrote:

> On Wed, Jul 15, 2009 at 3:37 PM, Frederik Ramm  
> wrote:
>> True. Everyone can delete all of London anyway. This doesn't mean  
>> that we
>> have a button on the web page that says: "Delete all of London".

As Pieren says, a revert is just another edit.  Perhaps this is a good  
solution:
   o When somebody hits the "revert" button (and is asked for a  
changeset comment), that sends email to the original submittor of the  
changeset notifying them that their changeset was reverted.
   o Revert everything that can be reverted.
   o When there are editing conflicts, include a link to Potlatch with  
a bbox for the conflict.

--
Russ Nelson - http://community.cloudmade.com/blog - 
http://wiki.openstreetmap.org/wiki/User:RussNelson
r...@cloudmade.com - Twitter: Russ_OSM - 
http://openstreetmap.org/user/RussNelson


___
talk mailing list
talk@openstreetmap.org
http://lists.openstreetmap.org/listinfo/talk


Re: [OSM-talk] Undo request button for changesets

2009-07-15 Thread Frederik Ramm
Hi,

Yann Coupin wrote:
> I completely agree with Pieren here. Unless you're part of the happy  
> fews, if case of vandalism/error, you're forced to painstakingly  
> repair by hand the problems

That is true, but we also have to look at the potential for problems.

("If everyone had a gun, it would be much easier for everyone to defend 
themselves, and the streets would thus become much safer for everyone." 
- Can you spot the problem?)

Often it is not one evil vandal against the poor mapper who repairs 
things. It is something like this: User A thinks that all cycleways 
should be highway=path,cycle=designated, and changes them all in his 
village. User B dislikes the idea and reverts the change (along with a 
small number of more usefule edits that A did and that were in the same 
changeset). User A reverts the revert. User B becomes angry and reverts 
not only the reversion of the revert, but all other edits that A did in 
the last week. User A asks his friend C for help, and C reverts 
everything that B ever did. B finds out that C is from Corse and deletes 
the whole island... etc. etc.

If we make these things too easy, then they will get abused. While it is 
true that the above scenario could happen today, it would require much 
more effort on the part of all parties, and this is a certain hurdle.

I am all for for giving more power to everyone but we have to think 
about how we can do this the right way.

Also, it is quite a challenge technically as well. A good revert system 
would have to visualise what it is about to revert, and would have to 
check in advance whether there are going to be any follow-on problems. A 
good revert system could even alert the user whose change is about to be 
reverted, and the user would have a chance to complain. It is surely 
something that we must think about, and surely not a matter of simply 
putting a button there that says "revert".

Bye
Frederik


___
talk mailing list
talk@openstreetmap.org
http://lists.openstreetmap.org/listinfo/talk


Re: [OSM-talk] Undo request button for changesets

2009-07-15 Thread Pieren
On Wed, Jul 15, 2009 at 3:37 PM, Frederik Ramm wrote:
> True. Everyone can delete all of London anyway. This doesn't mean that we
> have a button on the web page that says: "Delete all of London".
>

I agree that the button shouldn't be in front page where someone could
click on it by accident. It could be in the changset history for
instance.

I also agree that it might fail if we have conflicts because it took
too long time between the vandalism and its detection. But at least,
it would give a chance to revert if we react fast enough.

If I compare with Wikipedia, it would be like that anyone is allowed
to delete or modify a complete wiki page but other contributors (not
the last editor) are just allowed to restore the page line by line or
word by word...
Pieren

___
talk mailing list
talk@openstreetmap.org
http://lists.openstreetmap.org/listinfo/talk


Re: [OSM-talk] Undo request button for changesets

2009-07-15 Thread Bernhard zwischenbrugger
hi

I called it an "undo *request* button.
If I see a problem in the changelog I'm not able to revert it.
There is no defined way to report the problem.

The real problem starts if people start to repair manually.

What about a button that opens a dialog where people can describe the 
problem.
The problem description including changeset id could be sent to a 
special (ninja mapper) mailing  list, or to an RSS file, or to some 
special users,...

In some cases the revert could be done without human review.
Example:
o user wants to revert the own edits
o "power user" (mapper since 2 years) reverts changes of new user
o ...

Once the reporting UI is done it can start with manual reverts and can 
be improved doing automatic reverts.

Bernhard


> You are confusing what is technically possible with what is socially 
> acceptable. Frederik's point was that when you are doing a revert you 
> should consider whether (socially speaking) it is appropriate.
>
> So, to use his examples, if somebody mails him saying "I just made a 
> mistake, can you revert my edit" then that is obviously fine; but if 
> somebody on the other side of the world emails him saying "X keeps 
> making bad edits, can you revert them" then he will decline as he has no 
> way of knowing the background to the situation or who is right.
>
>   


___
talk mailing list
talk@openstreetmap.org
http://lists.openstreetmap.org/listinfo/talk


Re: [OSM-talk] Undo request button for changesets

2009-07-15 Thread Frederik Ramm
Hi,

Pieren wrote:
> It is a technical discussion because everybody can revert changes anyway.

True. Everyone can delete all of London anyway. This doesn't mean that 
we have a button on the web page that says: "Delete all of London".

(If we had a "revert" button next to a changeset, and I agree with the 
cautious voices who said that this is quite unlikely to work once a few 
hours or days have passed, then I would probably enable that button only 
for the user who committed the original change, and anyone else who 
clicks the button would just send a message to that user saying "please 
revert your change".)

Bye
Frederik

___
talk mailing list
talk@openstreetmap.org
http://lists.openstreetmap.org/listinfo/talk


Re: [OSM-talk] Undo request button for changesets

2009-07-15 Thread Yann Coupin
I completely agree with Pieren here. Unless you're part of the happy  
fews, if case of vandalism/error, you're forced to painstakingly  
repair by hand the problems (and that's if you're lucky and potlatch  
succeed at restoring all that's necessary). This is a problem I've  
personally encountered when a new user damaged a lot of stuff near  
where I live. Hopefully I could spot it quickly (thanks to itoworlds'  
graphical diff), but it was long to repair and I may very well have  
left some crumbs behind me as being thorough is hard when reverting  
way by way, node by node...

We really need an easy way to revert changes, and eventually combine  
that with local comities of "power users". Being open doesn't mean we  
have to remain all equals (esp. since, in fact, we're not).

Yann

Le 15 juil. 09 à 15:06, Pieren a écrit :

> It is a technical discussion because everybody can revert changes  
> anyway.
> The point is that anyone can delete 1000 buildings in your city in
> three clicks on JOSM but only few privileged people can revert this
> vandalism quickly (the ones how know the appropriate script). The
> others have to repair it manually 1000 times. It is a way to create a
> special category of "privileged contributors allowed to revert
> changesets" without saying it.


___
talk mailing list
talk@openstreetmap.org
http://lists.openstreetmap.org/listinfo/talk


Re: [OSM-talk] Undo request button for changesets

2009-07-15 Thread Pieren
On Wed, Jul 15, 2009 at 2:30 PM, Tom Hughes wrote:
> You are confusing what is technically possible with what is socially
> acceptable. Frederik's point was that when you are doing a revert you
> should consider whether (socially speaking) it is appropriate.
>
> So, to use his examples, if somebody mails him saying "I just made a
> mistake, can you revert my edit" then that is obviously fine; but if
> somebody on the other side of the world emails him saying "X keeps
> making bad edits, can you revert them" then he will decline as he has no
> way of knowing the background to the situation or who is right.
>
> Tom

It is a technical discussion because everybody can revert changes anyway.
The point is that anyone can delete 1000 buildings in your city in
three clicks on JOSM but only few privileged people can revert this
vandalism quickly (the ones how know the appropriate script). The
others have to repair it manually 1000 times. It is a way to create a
special category of "privileged contributors allowed to revert
changesets" without saying it.
Pieren

___
talk mailing list
talk@openstreetmap.org
http://lists.openstreetmap.org/listinfo/talk


Re: [OSM-talk] Undo request button for changesets

2009-07-15 Thread Tom Hughes
On 15/07/09 13:13, marcus.wolsc...@googlemail.com wrote:

> On Tue, 14 Jul 2009 08:49:11 +0100, Tom Hughes  wrote:
 >
>> Anyway, to draw on Frederik's workshop at SOTM the actual revert is in
>> many ways the easy part - the hard thing is establishing the authority
>> to do the revert. In other words the question of who gets to decide that
>> an edit is "bad" and should be reverted.
>
> A revert is itself just a new changeset.
> It is not a deletion of an existing changeset.
> So any user has the rights to do a revert
> if and only of that user has the right to
> make an edit.

You are confusing what is technically possible with what is socially 
acceptable. Frederik's point was that when you are doing a revert you 
should consider whether (socially speaking) it is appropriate.

So, to use his examples, if somebody mails him saying "I just made a 
mistake, can you revert my edit" then that is obviously fine; but if 
somebody on the other side of the world emails him saying "X keeps 
making bad edits, can you revert them" then he will decline as he has no 
way of knowing the background to the situation or who is right.

Tom

-- 
Tom Hughes (t...@compton.nu)
http://www.compton.nu/

___
talk mailing list
talk@openstreetmap.org
http://lists.openstreetmap.org/listinfo/talk


Re: [OSM-talk] Undo request button for changesets

2009-07-15 Thread marcus.wolschon
On Tue, 14 Jul 2009 08:49:11 +0100, Tom Hughes  wrote:
> Anyway, to draw on Frederik's workshop at SOTM the actual revert is in 
> many ways the easy part - the hard thing is establishing the authority 
> to do the revert. In other words the question of who gets to decide that 
> an edit is "bad" and should be reverted.


A revert is itself just a new changeset.
It is not a deletion of an existing changeset.
So any user has the rights to do a revert
if and only of that user has the right to
make an edit.

Same semantics as a revert in Wikpedia
(as long as a complete revert is technically
 possible).

Marcus

___
talk mailing list
talk@openstreetmap.org
http://lists.openstreetmap.org/listinfo/talk


Re: [OSM-talk] Undo request button for changesets

2009-07-14 Thread Yann Coupin
What would also be convenient would be a way to display older version  
in the browse history page. Take this way for instance:


http://www.openstreetmap.org/browse/way/23820548/history

The map on the top right overlays the current version on top of the  
tiles. But what if I want to see how it looked as version 4. I can  
click changeset 1793367 but it won't help, it would be great if the  
version was a link that changed the map to display the previous  
version. I understand that it would require to also get the nodes as  
they were at the time the changeset was created, but that would be a  
step in the right direction.


Also sometimes reverting a full changeset is more than needed,  
reverting a single object is all you want, and while it's possible by  
downloading an older version and modifying the XML by hand, it's  
certainly time consuming, and like others have said, reverting change  
should not take a lot more time than creating change.


Yann

Le 14 juil. 09 à 20:52, MP a écrit :


I think the worst case to revert is mass deletion - only Potlatch have
some facilities to find deleted lines given a BBOX, and only lines,
not POIs or relations. Such deletion is quite hard to spot.

As for reverting changesets - there is list of nodes/ways/relations
that was either added, changed or deleted in the changeset XML, so for
changed stuff, you can retrieve the previous version from history
(using the version from changeset as the "id of old version"), for new
stuff, you add action="delete" into the XML and for deleted stuff, you
recreate the last version (could be a bit tricky for deleted ways with
nodes, as you'll have to mangle the ID's a bit ...)

It shouldn't be probably hard to write such a thing ... perhaps some
of existing tools can be just a little adapted to produce this.

Load this into JOSM, use the conflict resolution in there to sort
things out (if someone tries to fix the thing manually) and upload
. changeset effectively reverted.


___
talk mailing list
talk@openstreetmap.org
http://lists.openstreetmap.org/listinfo/talk


Re: [OSM-talk] Undo request button for changesets

2009-07-14 Thread MP
I think the worst case to revert is mass deletion - only Potlatch have
some facilities to find deleted lines given a BBOX, and only lines,
not POIs or relations. Such deletion is quite hard to spot.

As for reverting changesets - there is list of nodes/ways/relations
that was either added, changed or deleted in the changeset XML, so for
changed stuff, you can retrieve the previous version from history
(using the version from changeset as the "id of old version"), for new
stuff, you add action="delete" into the XML and for deleted stuff, you
recreate the last version (could be a bit tricky for deleted ways with
nodes, as you'll have to mangle the ID's a bit ...)

It shouldn't be probably hard to write such a thing ... perhaps some
of existing tools can be just a little adapted to produce this.

Load this into JOSM, use the conflict resolution in there to sort
things out (if someone tries to fix the thing manually) and upload
. changeset effectively reverted.

Martin

___
talk mailing list
talk@openstreetmap.org
http://lists.openstreetmap.org/listinfo/talk


Re: [OSM-talk] Undo request button for changesets

2009-07-14 Thread Shaun McDonald


On 14 Jul 2009, at 18:13, John Smith wrote:



--- On Tue, 14/7/09, Chris Hunter  wrote:


Take a look at the history for this stretch of I-75 near
the TN/GA border, since I was unaware of the undo process, I
ended up spending 6+ hours redrawing I-75 (admittedly the
work needed to be done to fix TIGER, but still) because of a
single-keystroke mistake that the CTRL-Z undo in Potlatch
didn't allow me to undo.


Know that feeling, I had a way with > 4000 points I was trying to  
split and at the time JOSM and potlatch only made a mess of it  
leaving all the nodes and no connecting way. Because I didn't know I  
could undo a changeset I spent hours reconnecting all the nodes by  
hand.





In that case the revert wouldn't work, because you can't save any way  
with more than 2000 nodes (even if it is part of a revert).


Shaun



smime.p7s
Description: S/MIME cryptographic signature
___
talk mailing list
talk@openstreetmap.org
http://lists.openstreetmap.org/listinfo/talk


Re: [OSM-talk] Undo request button for changesets

2009-07-14 Thread John Smith

--- On Tue, 14/7/09, Chris Hunter  wrote:

> Take a look at the history for this stretch of I-75 near
> the TN/GA border, since I was unaware of the undo process, I
> ended up spending 6+ hours redrawing I-75 (admittedly the
> work needed to be done to fix TIGER, but still) because of a
> single-keystroke mistake that the CTRL-Z undo in Potlatch
> didn't allow me to undo.

Know that feeling, I had a way with > 4000 points I was trying to split and at 
the time JOSM and potlatch only made a mess of it leaving all the nodes and no 
connecting way. Because I didn't know I could undo a changeset I spent hours 
reconnecting all the nodes by hand.


  

___
talk mailing list
talk@openstreetmap.org
http://lists.openstreetmap.org/listinfo/talk


Re: [OSM-talk] Undo request button for changesets

2009-07-14 Thread Richard Fairhurst

Chris Hunter wrote:
> Take a look at the history for this stretch of I-75 near the TN/GA 
> border, since I was unaware of the undo process, I ended up spending 
> 6+ hours redrawing I-75 (admittedly the work needed to be done 
> to fix TIGER, but still) because of a single-keystroke mistake 
> that the CTRL-Z undo in Potlatch didn't allow me to undo.

a) what was the mistake?
b) are you aware that Potlatch already has a history/revert function?

cheers
Richard
-- 
View this message in context: 
http://www.nabble.com/Undo-request-button-for-changesets-tp24474935p24483893.html
Sent from the OpenStreetMap - General mailing list archive at Nabble.com.


___
talk mailing list
talk@openstreetmap.org
http://lists.openstreetmap.org/listinfo/talk


Re: [OSM-talk] Undo request button for changesets

2009-07-14 Thread Chris Hunter
For those of us who live in areas that are pretty sparse on mappers, I don't
see this being a problem.  Speaking from personal experience, just seeing an
undo button on your user history would prevent a lot of conflicts from
people trying to "fix" their mistakes.

Take a look at the history for this stretch of I-75 near the TN/GA border,
since I was unaware of the undo process, I ended up spending 6+ hours
redrawing I-75 (admittedly the work needed to be done to fix TIGER, but
still) because of a single-keystroke mistake that the CTRL-Z undo in
Potlatch didn't allow me to undo.

http://www.openstreetmap.org/?minlon=-85.6267052&minlat=34.8502047&maxlon=-84.2608284&maxlat=35.0374212&box=yes

Chris

On Tue, Jul 14, 2009 at 12:39 PM,  wrote:

> As soon as someone has edited anything in the changeset after the version
> in the changeset, then you have a conflict and can't do a simple one click
> revert. OpenStreetMap data is a *lot* more complex than the Wikipedia when
> it comes to reverting due to the referential integrity. That is key to there
> not being a simple revert system for whole changesets.
>
>
>> If the edit can't be reverted because of conflicts, simply show a:
>> "can't revert because of editing conflicts" message.
>>
>> I guess this would already solve 99.99% of all cases.
>>
>
> Only in the seconds/hours after the edit was made (depending on how active
> the mappers in the area are). The further back in time you go the more
> conflicts you get and the harder it is.
>
> Shaun
>
___
talk mailing list
talk@openstreetmap.org
http://lists.openstreetmap.org/listinfo/talk


Re: [OSM-talk] Undo request button for changesets

2009-07-14 Thread Renaud MICHEL
Le mardi 14 juillet 2009 à 13:13, Francois Van Der Biest a écrit :
> Therefore, I imagine we should show features which have been
> created/updated/deleted in different colors (eg: green/blue/red) on
> the changeset page.

+1
That would be really cool.
I've been missing such a feature since changeset have been introduced.

-- 
Renaud Michel


___
talk mailing list
talk@openstreetmap.org
http://lists.openstreetmap.org/listinfo/talk


Re: [OSM-talk] Undo request button for changesets

2009-07-14 Thread Ulf Lamping
Shaun McDonald schrieb:
> 
> On 14 Jul 2009, at 09:50, Ulf Lamping wrote:
> 
>> Tom Hughes schrieb:
>>
>>> because in general terms it won't work - reverting
>>> will often need manual intervention to resolve conflicts.
>>
>> I've heard this argument many times before, but no prove that it is
>> actually true. Why not do it the osm way, implement it the simplest
>> possible way and see how far we get?
> 
> People who have reverted stuff in the past have already come across the 
> conflicts. This is especially true when you have someone who has tried 
> to fix the problem manually before asking for help.

True.

But: IMHO we would probably have a lot less of those manual fixes, when 
there would be an easier way to "repair stuff" like a revert button.

A normal mapper currently simply don't have a lot of other options to 
repair stuff than to try to do it manually :-)

> As soon as someone has edited anything in the changeset after the 
> version in the changeset, then you have a conflict and can't do a simple 
> one click revert. OpenStreetMap data is a *lot* more complex than the 
> Wikipedia when it comes to reverting due to the referential integrity. 
> That is key to there not being a simple revert system for whole changesets.

Yes, I'm aware of that. However, the question is how often will this 
actually happen if someone wants to revert stuff - IMO this is only 
rarely the case for the time period where a revert is really interesting.

>> If the edit can't be reverted because of conflicts, simply show a:
>> "can't revert because of editing conflicts" message.
>>
>> I guess this would already solve 99.99% of all cases.
> 
> Only in the seconds/hours after the edit was made (depending on how 
> active the mappers in the area are). The further back in time you go the 
> more conflicts you get and the harder it is.

Yes, the revert will probably only be possible in the few hours / days 
after the changeset was done.

I don't expect a changeset done three months ago to be actually 
revertable - probably a good thing in that case ;-)

Regards, ULFL

___
talk mailing list
talk@openstreetmap.org
http://lists.openstreetmap.org/listinfo/talk


Re: [OSM-talk] Undo request button for changesets

2009-07-14 Thread Francois Van Der Biest
I agree with Pieren on this.

Monitoring changesets should also be as easy as in Wikipedia.
Therefore, I imagine we should show features which have been
created/updated/deleted in different colors (eg: green/blue/red) on
the changeset page.
As this could result in a big vector load for the browser, I guess
this would have to be a static raster tile, which could be loaded as
an OpenLayers.Layer.Image in OpenLayers.

F.

On Tue, Jul 14, 2009 at 12:47 PM, Pieren wrote:
> On Tue, Jul 14, 2009 at 12:31 PM, Chris Hill wrote:
>> And when a vandal chooses to undo all the changesets you have added in the
>> last few months how will you feel then?
>>
>> Cheers, Chris
>>
>
> Then I will revert his changeset. And if it happens again, I will call
> the community for some help. As wikipedian do every day in one click.
> We cannot, in one way, make edition easy and quick (e.g. the online
> editor) and in the other way, make reverting edits complicated and
> long. This could work only if you give access to educated and trusted
> contributors.
> Pieren

___
talk mailing list
talk@openstreetmap.org
http://lists.openstreetmap.org/listinfo/talk


Re: [OSM-talk] Undo request button for changesets

2009-07-14 Thread Shaun McDonald


On 14 Jul 2009, at 09:50, Ulf Lamping wrote:


Tom Hughes schrieb:


because in general terms it won't work - reverting
will often need manual intervention to resolve conflicts.


I've heard this argument many times before, but no prove that it is
actually true. Why not do it the osm way, implement it the simplest
possible way and see how far we get?


People who have reverted stuff in the past have already come across  
the conflicts. This is especially true when you have someone who has  
tried to fix the problem manually before asking for help.


As soon as someone has edited anything in the changeset after the  
version in the changeset, then you have a conflict and can't do a  
simple one click revert. OpenStreetMap data is a *lot* more complex  
than the Wikipedia when it comes to reverting due to the referential  
integrity. That is key to there not being a simple revert system for  
whole changesets.




If the edit can't be reverted because of conflicts, simply show a:
"can't revert because of editing conflicts" message.

I guess this would already solve 99.99% of all cases.


Only in the seconds/hours after the edit was made (depending on how  
active the mappers in the area are). The further back in time you go  
the more conflicts you get and the harder it is.


Shaun

smime.p7s
Description: S/MIME cryptographic signature
___
talk mailing list
talk@openstreetmap.org
http://lists.openstreetmap.org/listinfo/talk


Re: [OSM-talk] Undo request button for changesets

2009-07-14 Thread Pieren
On Tue, Jul 14, 2009 at 12:31 PM, Chris Hill wrote:
> And when a vandal chooses to undo all the changesets you have added in the
> last few months how will you feel then?
>
> Cheers, Chris
>

Then I will revert his changeset. And if it happens again, I will call
the community for some help. As wikipedian do every day in one click.
We cannot, in one way, make edition easy and quick (e.g. the online
editor) and in the other way, make reverting edits complicated and
long. This could work only if you give access to educated and trusted
contributors.
Pieren

___
talk mailing list
talk@openstreetmap.org
http://lists.openstreetmap.org/listinfo/talk


Re: [OSM-talk] Undo request button for changesets

2009-07-14 Thread Chris Hill




And when a vandal chooses to
undo all the changesets you have added in the last few months how will
you feel then?

Cheers, Chris

Pieren wrote:

  On Tue, Jul 14, 2009 at 10:50 AM, Ulf Lamping wrote:
  
  
Why not? Wikipedia seems to be living pretty well with that approach
(yes, I'm aware that there are differences to osm, as the revert button
of a page only appears on that pages history, not on a "global revert
list" as it would be in our case).

  
  
I agree with Ulf. We should follow the wikipedia approach.
Undoing mistakes/vandalism should be as easy as creating mistakes/vandalism.

Pieren

___
talk mailing list
talk@openstreetmap.org
http://lists.openstreetmap.org/listinfo/talk

  




___
talk mailing list
talk@openstreetmap.org
http://lists.openstreetmap.org/listinfo/talk


Re: [OSM-talk] Undo request button for changesets

2009-07-14 Thread Pieren
On Tue, Jul 14, 2009 at 10:50 AM, Ulf Lamping wrote:
> Why not? Wikipedia seems to be living pretty well with that approach
> (yes, I'm aware that there are differences to osm, as the revert button
> of a page only appears on that pages history, not on a "global revert
> list" as it would be in our case).

I agree with Ulf. We should follow the wikipedia approach.
Undoing mistakes/vandalism should be as easy as creating mistakes/vandalism.

Pieren

___
talk mailing list
talk@openstreetmap.org
http://lists.openstreetmap.org/listinfo/talk


Re: [OSM-talk] Undo request button for changesets

2009-07-14 Thread John Smith

--- On Tue, 14/7/09, Tom Hughes  wrote:

> If you're saying that it should mail some mythical team of
> ninja mappers 
> who will spring into action and revert the changeset then
> you're going 
> to need to establish the team of ninja mappers first before
> we can add 
> the button.

Why not start simpler, just have a changeset flagged and will then appear on a 
secondary list that people can watch.


  

___
talk mailing list
talk@openstreetmap.org
http://lists.openstreetmap.org/listinfo/talk


Re: [OSM-talk] Undo request button for changesets

2009-07-14 Thread Ulf Lamping
Tom Hughes schrieb:
> On 14/07/09 08:30, Bernhard zwischenbrugger wrote:
> 
>> Now we have the changesets like
>> http://www.openstreetmap.org/browse/changeset/1815935
>>
>> It's relative easy to identify bad edits.
>>
>> Is it possible to add an "undo request button" or "spam button" to this
>> page?
> 
> What are you expecting this button to do?
> 
> If you want it to actually revert the change then the simple answer is 
> that such a button probably won't appear, both because we don't want to 
> make it too easy 

Why not? Wikipedia seems to be living pretty well with that approach 
(yes, I'm aware that there are differences to osm, as the revert button 
of a page only appears on that pages history, not on a "global revert 
list" as it would be in our case).

> because in general terms it won't work - reverting 
> will often need manual intervention to resolve conflicts.

I've heard this argument many times before, but no prove that it is 
actually true. Why not do it the osm way, implement it the simplest 
possible way and see how far we get?

If the edit can't be reverted because of conflicts, simply show a: 
"can't revert because of editing conflicts" message.

I guess this would already solve 99.99% of all cases.

> If you're saying that it should mail some mythical team of ninja mappers 
> who will spring into action and revert the changeset then you're going 
> to need to establish the team of ninja mappers first before we can add 
> the button.

Well, chicken and egg problem :-)

> Anyway, to draw on Frederik's workshop at SOTM the actual revert is in 
> many ways the easy part - the hard thing is establishing the authority 
> to do the revert. In other words the question of who gets to decide that 
> an edit is "bad" and should be reverted.

Yes, if we need to have those ninja mapper group, then this problem appears.

Regards, ULFL

___
talk mailing list
talk@openstreetmap.org
http://lists.openstreetmap.org/listinfo/talk


Re: [OSM-talk] Undo request button for changesets

2009-07-14 Thread Tom Hughes
On 14/07/09 08:30, Bernhard zwischenbrugger wrote:

> Now we have the changesets like
> http://www.openstreetmap.org/browse/changeset/1815935
>
> It's relative easy to identify bad edits.
>
> Is it possible to add an "undo request button" or "spam button" to this
> page?

What are you expecting this button to do?

If you want it to actually revert the change then the simple answer is 
that such a button probably won't appear, both because we don't want to 
make it too easy and because in general terms it won't work - reverting 
will often need manual intervention to resolve conflicts.

If you're saying that it should mail some mythical team of ninja mappers 
who will spring into action and revert the changeset then you're going 
to need to establish the team of ninja mappers first before we can add 
the button.

Anyway, to draw on Frederik's workshop at SOTM the actual revert is in 
many ways the easy part - the hard thing is establishing the authority 
to do the revert. In other words the question of who gets to decide that 
an edit is "bad" and should be reverted.

Tom

-- 
Tom Hughes (t...@compton.nu)
http://www.compton.nu/

___
talk mailing list
talk@openstreetmap.org
http://lists.openstreetmap.org/listinfo/talk