Hello,
There was a project for integration of OpenStreetMap material in Wikimedia projects at http://cassini.toolserver.org/tile-browse/. Cassini seems to be down now. Is the OSM project abandoned or has it moved to another place?
Thanks Marcus Buck User:Slomox
Marcus Buck schrieb:
Hello,
There was a project for integration of OpenStreetMap material in Wikimedia projects at http://cassini.toolserver.org/tile-browse/. Cassini seems to be down now. Is the OSM project abandoned or has it moved to another place?
Cassini was just a test-server, no stable production environment. It has moved closer into the normal toolserver cluster. The map-styles we're rendering are viewable in http://toolserver.org/~osm/styles/.
We're planning to get all 270+ language maps back online on a production server but there's no schedule on this.
Peter
Peter Körner hett schreven:
Marcus Buck schrieb:
Hello,
There was a project for integration of OpenStreetMap material in Wikimedia projects at http://cassini.toolserver.org/tile-browse/. Cassini seems to be down now. Is the OSM project abandoned or has it moved to another place?
Cassini was just a test-server, no stable production environment. It has moved closer into the normal toolserver cluster. The map-styles we're rendering are viewable in http://toolserver.org/~osm/styles/.
We're planning to get all 270+ language maps back online on a production server but there's no schedule on this.
Thanks! Is there any information about what's the showstopper or how complex it is to get back all the languages (e.g. in manhours)?
Marcus Buck User:Slomox
Peter Körner hett schreven:
Marcus Buck schrieb:
Thanks! Is there any information about what's the showstopper or how complex it is to get back all the languages (e.g. in manhours)?
I can tell you about my showstopper and this is the wonderful weather.
Of course I have no right to request anything from you. I just would like to have the maps back - cause they are potentially very, very useful, especially for our smaller languages, for which there often are no maps available at all - and I'd like to know how good my chances are. How much work is it? Half an hour? A day 9am-5pm? A month 9am-5pm? I do not even have a vague idea about the problems that need to be solved. Perhaps somebody can shed some light on it?
Marcus Buck User:Slomox
Marcus Buck schrieb:
How much work is it? Half an hour? A day 9am-5pm? A month 9am-5pm? I do not even have a vague idea about the problems that need to be solved. Perhaps somebody can shed some light on it?
Setting up th styles it not an issue, let's say around 1 hour. Th think I'm not sure about is, if we should do it on our development / toolserver.
My original Idea was to have the toolserver there for the people to play around, test crazy or useful styles ans such.
The multilingual maps should go on the live system (what previously was cassini). But there have been no pretensions to set up cassini with an render stack. Maybe we just have to open a ticket in jira to get it done [1]. Once the system is set up, I'd go on importing another clone of the database [2] and set up the multilingual rendering again - and that's it.
Peter
[1] http://... [2] i'M unsure about this one. Should we use ptolemy for live rendering or should cassini carry it's own, local copy?
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1
Peter Körner:
The multilingual maps should go on the live system (what previously was cassini). But there have been no pretensions to set up cassini with an render stack.
cassini is not (and never was) the production renderer. It has been named to 'adenia' and is now serving Toolserver user databases.
WMF returned the hardware that was to be used for this to us, since it was sitting around unused and no one was working on it, and we have re-purposed it.
If anyone is interesting is integrating maps into Wikipedia, the first thing to do is probably to get WMF to buy some new hardware for it.
- river.
River Tarnell schrieb:
If anyone is interesting is integrating maps into Wikipedia, the first thing to do is probably to get WMF to buy some new hardware for it.
Actually, the first thing we need is a working, scalable, usable way to actually *use* these maps in wikipedia. Then it should be easy to get the wmf to supply the hardware we need. As far as I know, the maps extensions "is being worked on"...
-- daniel
I saw that cassini has been renamed to adenia and it seems it has a new purpose now, so I just asked to Maps-l first before opening the ticket but I forgot to remove the reference in this mail. Sorry for the confusion.
Peter
Peter Körner schrieb:
Marcus Buck schrieb:
How much work is it? Half an hour? A day 9am-5pm? A month 9am-5pm? I do not even have a vague idea about the problems that need to be solved. Perhaps somebody can shed some light on it?
Setting up th styles it not an issue, let's say around 1 hour. Th think I'm not sure about is, if we should do it on our development / toolserver.
My original Idea was to have the toolserver there for the people to play around, test crazy or useful styles ans such.
The multilingual maps should go on the live system (what previously was cassini). But there have been no pretensions to set up cassini with an render stack. Maybe we just have to open a ticket in jira to get it done [1]. Once the system is set up, I'd go on importing another clone of the database [2] and set up the multilingual rendering again - and that's it.
Peter
[1] http://... [2] i'M unsure about this one. Should we use ptolemy for live rendering or should cassini carry it's own, local copy?
Maps-l mailing list Maps-l@lists.wikimedia.org https://lists.wikimedia.org/mailman/listinfo/maps-l
Peter Körner schrieb:
Marcus Buck schrieb:
How much work is it? Half an hour? A day 9am-5pm? A month 9am-5pm? I do not even have a vague idea about the problems that need to be solved. Perhaps somebody can shed some light on it?
Setting up th styles it not an issue, let's say around 1 hour. Th think I'm not sure about is, if we should do it on our development / toolserver.
I started a testcase with language overlays instead of complete language maps. This should save a lot of RAM (loading all 270+ map styles took ~16GB of Memory only for renderd).
It's still rendering the low-zoom tiles so it is very slow. I also activated only three language overlays to test the setup but the other stlye-files are already in place.
So feel free to play around: http://toolserver.org/~osm/locale/
Peter
Peter Körner hett schreven:
Peter Körner schrieb:
Marcus Buck schrieb:
How much work is it? Half an hour? A day 9am-5pm? A month 9am-5pm? I do not even have a vague idea about the problems that need to be solved. Perhaps somebody can shed some light on it?
Setting up th styles it not an issue, let's say around 1 hour. Th think I'm not sure about is, if we should do it on our development / toolserver.
I started a testcase with language overlays instead of complete language maps. This should save a lot of RAM (loading all 270+ map styles took ~16GB of Memory only for renderd).
It's still rendering the low-zoom tiles so it is very slow. I also activated only three language overlays to test the setup but the other stlye-files are already in place.
So feel free to play around: http://toolserver.org/~osm/locale/
Great! Thanks!
Is there any way to distinguish localized placenames and unlocalized placenames (as an option)? Like unlocalized names being displayed in a different color. That would make it easier for maintainers to improve the OSM data.
Marcus Buck User:Slomox
Is there any way to distinguish localized placenames and unlocalized placenames (as an option)? Like unlocalized names being displayed in a different color. That would make it easier for maintainers to improve the OSM data.
Atm. there is none, but I could think of another style for this but I don't want another 270+ styles on the server. I could think of adding such styles for special, selected languages, but I can also imagine that a none-map-tool (like [1]) or a openlayers-vector-map could do ab better job.
Peter
[1] http://toolserver.org/~mazder/multilingual-country-list/
Дана Wednesday 16 June 2010 12:07:19 Peter Körner написа:
I started a testcase with language overlays instead of complete language maps. This should save a lot of RAM (loading all 270+ map styles took ~16GB of Memory only for renderd).
It's still rendering the low-zoom tiles so it is very slow. I also activated only three language overlays to test the setup but the other stlye-files are already in place.
So feel free to play around: http://toolserver.org/~osm/locale/
Works great! :) Could you add this to the local CSS?
.dataLayersDiv { height: 20em; margin-right: 1em; overflow: scroll; }
because the overlay list is too long and some of the languages are off the screen.
Not sure whether it is appropriate for this list, but I was thinking on how to properly cascade languages, without coming to a full conclusion.
First, if I am watching Serbian layer, and there is place that doesn't have Serbian name, but does have Croatian name, I would prefer to see Croatian name instead of the "main" name. There probably are more such cases (Nynorsk/Bokmal?). (I see you are not rendering those but I guess you will eventually.)
A harder problem is that for some uses, users might legitimately need to display several languages. For example, there is this square in Belgrade that has:
lang=Трг републике lang:sr@Latn=Trg republike lang:en=Republic Square
A tourist from USA, viewing a map of Belgrade, will want to have all three: the first is actual name of the square as actually written on the street signs on the square (which he needs so that he could read the square name on the signs and verify that he really is there); the second is transcription of the name to Latin alphabet (which he needs so that he could know how to read the square name aloud so that he could ask the locals how to get there); the third is the English name of the square (which he needs to ask the English-speaking locals how to get there).
I am not sure how to display all of this without drowning the map with text, and I am not sure how people would choose what they want to see (maybe someone who is not traveling to Belgrade but only looking at the map would prefer to see only sr@Latn and English?).
If this is implemented, next step is automated transcription...
Nikola Smolenski schrieb:
Дана Wednesday 16 June 2010 12:07:19 Peter Körner написа: Works great! :) Could you add this to the local CSS?
.dataLayersDiv { height: 20em; margin-right: 1em; overflow: scroll; }
because the overlay list is too long and some of the languages are off the screen.
Yes yesterday evening I saw that on my notebook screen, too. I added the snippet but used 30em as height. Is this ok for you?
Not sure whether it is appropriate for this list, but I was thinking on how to properly cascade languages, without coming to a full conclusion.
First, if I am watching Serbian layer, and there is place that doesn't have Serbian name, but does have Croatian name, I would prefer to see Croatian name instead of the "main" name. There probably are more such cases (Nynorsk/Bokmal?). (I see you are not rendering those but I guess you will eventually.)
When a country has two or more used languages, I could think of having a layer that chooses the first available name-tag, but if it's only a personal preference, we need to find another way. This is also true for the next case.
A harder problem is that for some uses, users might legitimately need to display several languages. For example ...
I don't think that this can be done on the server without sth. like a wms service. It's just not possible to list all possible combinations on the server, so they need to be a) rendered on-demand on the server b) rendered on the client
b) is the most interesting one. I talked to a colleague about doing this with an svg overlay but most browsers font capabilities does not come close to what agg does for mapnik, so it will look horrible.
if someone with wms experience wants to play around with a), that could be a real killer service.
Peter
On 06/17/2010 08:55 AM, Peter Körner wrote:
Nikola Smolenski schrieb:
Дана Wednesday 16 June 2010 12:07:19 Peter Körner написа: Works great! :) Could you add this to the local CSS?
.dataLayersDiv { height: 20em; margin-right: 1em; overflow: scroll; }
because the overlay list is too long and some of the languages are off the screen.
Yes yesterday evening I saw that on my notebook screen, too. I added the snippet but used 30em as height. Is this ok for you?
It is OK for me.
A harder problem is that for some uses, users might legitimately need to display several languages. For example ...
I don't think that this can be done on the server without sth. like a wms service. It's just not possible to list all possible combinations on the server, so they need to be
Of course, it would be ludicrous to try having all possible combinations, but I believe a few of the most common ones could be aimed for (for example, for every language display only language, or local language + that language).
On 06/17/2010 09:28 AM, Nikola Smolenski wrote:
On 06/17/2010 08:55 AM, Peter Körner wrote:
Nikola Smolenski schrieb:
Дана Wednesday 16 June 2010 12:07:19 Peter Körner написа: Works great! :) Could you add this to the local CSS?
.dataLayersDiv { height: 20em; margin-right: 1em; overflow: scroll; }
because the overlay list is too long and some of the languages are off the screen.
Yes yesterday evening I saw that on my notebook screen, too. I added the snippet but used 30em as height. Is this ok for you?
It is OK for me.
Except that margin-right: 0.2em produces a scroll bar that is partially obscured by the right edge of the browser window.
toolserver-l@lists.wikimedia.org