Future of DeepEarth?

Apr 4, 2009 at 11:41 PM
Now that the Microsoft Silverlight map control has been released, is there a future for the DeepEarth project ? Personally I think there is, but I'm sure it will be a question that quite a few people who have been following DeepEarth's development over the past few months will be asking themselves now.

Should we gather together a list of strengths / weaknesses of DeepEarth compared to Microsoft's control (and perhaps also ESRI's and the others out there) and have that on the codeplex doc pages ?
Developer
Apr 5, 2009 at 1:22 AM
I have not yet looked at the MS control yet but I assume that it only uses VE tiles
to get a account at VE will cost quite a bit of money from what I have read
 
the difference between DepEarth and the MS control is, DeepEarth is made to work with many different Tile Providers and then offer the developer a common way to deal with them all regardless of who is providing the tiles
 
So I guess there will still be people who will like some of the options DeepEath offers..?
Coordinator
Apr 5, 2009 at 12:13 PM
Edited Apr 5, 2009 at 12:15 PM
Hey guys, I've been lurking for a week after getting back from Egypt. The simple answer is, yes there is a clear future for DeepEarth.

One of the biggest issues we faced with DeepEarth was support / SLA etc for commerical applications. The reality is that we don't have that as an enthusist control, however we do have tonnes to offer.
The Microsoft control has a couple of nice features like world wrap, zoomed panning between points etc. But it is still a very basic mapping control, the work we have been doing with SQL 2008 integration and the drawing tools from noobie are not only well above what will be offered in that control but are actually (with slight modification) compatible with that control.

I see the future of DeepEarth as two fold:

1) A set of reusable controls. This actually can include tile layers, vector layers as well as the obvious nav bar, scalebar, drawing tools and panels.
2) An open source rapid development enviroment for new cutting edge functionality. There are times when you are going to want to tweak the underlying MSI or canvas to do something clever. I'm certainly keen to upgrade to Silverlight 3, not that the MSFT control can't be used in SL3 but is only Silverlight 2.

And of course DeepEarth is free!

It may interest you to know that although the MSFT control doesn't have one of our geomentry rendering issues, it does have the one where the fill flips around and the polygon dissapears entirely if you zoom into a large polygon's centre. DeepEarth's performance for more geometry is also better in my early tests (I had to drastically reduce the complexity of polygons in one demo that was working fine in DeepEarth)

I kick off a migration of one of my clients apps from DeepEarth to the Microsoft control on Tuesday, happy to report on how that goes. 
John.
Apr 5, 2009 at 2:32 PM

Hi

“the work we have been doing with SQL 2008 integration “

I think this is a big missing piece. If DeepEarth can provide this along with the very nice drawing tools and panel controls already in place, it will be well ahead of the MS VE MapControl.

I look forward to playing with a real datastore. Silverlight communication with the outside world has seemed a bit complicated to me. I’ve noticed that the Linq stuff has been downplayed by MS recently, which may make it an iffy approach. As far as I can tell the current Silverlight approach is to setup an .asmx wrapper for database calls and then pull the results into the Silverlight side where the code behind can add it to the view. I’m from a Java background so this seems similar to using servlets to pass jdbc query data into a javascript viewing client.

Even with the new spatial capability, MS SQL it is lacking a lot of the features of PostgreSQL/PostGIS. So …  it would be nice to have integration through OGC WFS. That way the backend data store is not restricted to MS for vector data as well as tiles. KML and GeoRSS import capability are nice for that reason as well. If DeepEarth can broaden import support to OGC servers for vectors it would give it another advantage.

I gather that the VE license may be adjusted in some manner with the final release of Silverlight VE Map Control. This may also help DeepEarth since the hints I’ve noticed have to do with the transaction tile accounting.

DeepEarth holds an advantage on multiple tile sources so developers can offer low end versions of a web app using OSM with a price scaling against multiple backend tile source choices. I’ve been able to do the same with Silverlight VE MapControl by connecting alternate tile sources and turning off the VE tiles. Even though this would drop transaction costs to nil, I imagine there will be some license restriction doing that for a commercial client. http://www.cadmaps.com/gisblog/?p=54

As far as I can tell with the Silverlight VE MapControl you will be on your own connecting out to a database. Unlike DeepEarth, drawing tools, beautiful panels, etc would need to be implemented as well.

rkgeorge

From: soulsolutions [mailto:notifications@codeplex.com]
Sent: Sunday, April 05, 2009 6:14 AM
To: rkgeorge@cadmaps.com
Subject: Re: Future of DeepEarth? [deepearth:52293]

From: soulsolutions

Hey guys, I've been lurking for a week after getting back from Egypt. The simple answer is, yes there is a clear future for DeepEarth.

One of the biggest issues we faced with DeepEarth was support / SLA etc for commerical applications. The reality is that we don't have that as an enthusist control, however we do have tonnes to offer.
The Microsoft control has a couple of nice features like world wrap, zoomed panning between points etc. But it is still a very basic mapping control, the work we have been doing with SQL 2008 integration and the drawing tools from noobie are not only well above what will be offered in that control but are actually (with slight modification) compatible with that control.

I see the future of DeepEarth as two fold:

1) A set of reusable controls. This actually can invlude tile layers, vector layers as well as the obvious nav bar, scalebar, drawing tools and panels.
2) An open source rapid development enviroment for new cutting edge functionality. There are times when you are going to want to tweak the underlying MSI or canvas to do something clever. I'm certainly keen to upgrade to Silverlight 3, not that the MSFT control can be used in SL3 but is only Silverlight 2.

And of course DeepEarth is free!

It may interest you to know that although the MSFT control doesn't have one of our geomentry rendering issues, it does have the one where the fill flips around and the polygon dissapears entirely if you zoom into a large polygon's centre. DeepEarth's performance for more geometry is also better in my early tests (I had to drastically reduce the complexity of polygons in one demo that was working fine in DeepEarth)

I kick off a migration of one of my clients apps from DeepEarth to the Microsoft control on Tuesday, happy to report on how that goes.
John.

Read the full discussion online.

To add a post to this discussion, reply to this email (deepearth@discussions.codeplex.com)

To start a new discussion for this project, email deepearth@discussions.codeplex.com

You are receiving this email because you subscribed to this discussion on CodePlex. You can unsubscribe or change your settings on codePlex.com.

Please note: Images and attachments will be removed from emails. Any posts to this discussion will also be available online at codeplex.com

Apr 10, 2009 at 12:46 PM
If I had to start over agaian, and if the MS control was released for production, I would start there.I don't want to be rude and I really want to thank all of you for the tremendous work you've been doing. But...

What is driving me nuts in DeepEarth is the zig-zag course the architecture is following. We have built a beauttifull site which would never ever would have been possible without your work. The site is running against DeepEarth 1.0 with some tweeks and subclassing. We are doing nothing special. An aerial view of the world as "wallpaper" and several layers with large amounts of polygons, polylines and some images. No realy fancy graphics, no routing info. We just need the cpu/gpu power of Silverlight to get this done.

While building this had I had change the  code over and over again to adapt to the changining architecture of DE. Now everything is stable against 1.0. I made an attempt to upgrade to DE 1.1. After changing the code (again..) I could not get our layers to show. Can't see why. We'll stick to 1.0

The promissing thing with the MS control is the clear simple API it is presenting. The posts by Chris Pietschmann, like  http://pietschsoft.com/post/2009/03/Virtual-Earth-Silverlight-Using-MouseClick-Event-To-Add-Pushpins.aspx, describe all we need. Including a way to work without the expensive MS VE service.

Coordinator
Apr 11, 2009 at 12:16 AM
Peter,
I understand your pain, this project is all about people learning silverlight and contributing to something that becomes bigger and more useful then what we can do seperately. Unlike some projects their is no company behind this, there is 5 or 6 people who have contributed hundreds of hours of their time.
I'm halfway through converting a very complex application (many tile overlays, SQL 2008 points, lines and polygons, custom layer controls, noobie's drawing tools etc) and the structure isn't that different. I like some of the simplicity the MSFT control has in terms of simple using the AddChild() for everything and the flexibility but there are area's where DeepEarth outperforms or has extra core features.
I don't see how DeepEarth could compete with MSFT and ESRI on support and backwards compatibility issues, we don't have a staff of people and the funds to do that. What DeepEarth could do really well is innovation.
Keep in mind that DeepEarth uses the exact same VE service as the MSFT control does, simply in their CTP they have hard coded the Key for easier setup. When it moves out of CTP (you can not use a CTP in production) you will need to use the token system. We are all hoping the VE authentication system moves over to a simple appID style.

I'd love to hear other's views and thanks for the feedback Peter,
John.
Apr 14, 2009 at 10:01 AM
hi John,

yep SQL 2008 is great. We're using it too, and it is a great help. Out major pain with the product was the geo-awareness of the the MS team. In the first versions of the beta it was unclear what comes first: latitude or longitude. The first beta's did it the other way round geo people were used to. But I don't see what DE has to do with sql 2008 when it comes to integration. DE is running on the client and should have no awareness of a database server. All information comes to the client using services which should talk in geo types. These will have their counterparts in SQL 2008, but they can not be the same types.

I have not worked with the MS control yet but as the description reads it should be a breeze to use it without the dreaded MS tile server. http://pietschsoft.com/post/2009/03/Virtual-Earth-Silverlight-Using-Tile-Layers-to-Overlay-Custom-Map-Imagery.aspx You can set the base map to empty, the tile server will never be hit then. And you can easily overlay the map with a tileserver of preference. As said, I have not tried this yet. In case it (does not) work it will be worth a blogpost. I had planned to post more on DeepEarth. But my main problem is the moving target. The post I had done on the custom pushpins has become completely obsolete in a short time. And as said, trying to switch from 1.0 to 1.1 was a problem on itself.

I do understand DE is a project where people are learning and I really do appreceate the huge amounts of work they have done. But as said I do have great difficulties with the architecture. To name an example:
We are using a coordcontrol on our map but we are not satisfied with the way it displays the map position. Latitude and longitude are displayed as double's with a 4 digit precision behind the comma. lat and long are separated by another comma (in Dutch the decimal separator is comma). As a seaman I want my position in real coordinate. Like 45° 12' 20". Which takes some calculations and some formatting. But coordcontrol is completely closed to any change in the output format. Anything else than the default requires changing the source code. That's just bad.

Keep the discussion alive !

best

Peter