March 14, 2011 1 Comment
Well the 2011 Dev Summit wrapped up with a bunch of folks walking around with sore arms from the dodgeball tournament and you could tell Jim McKinney was disappointed his team wasn’t in the finals. As has been the pattern over the last 6 years, Esri raises the bar a little bit each year making this better and better. The sessions we better, they were nimble in moving rooms around to accommodate crowds, and the food was better. This is by far and away the best technical show they offer, and with the direct access to the product teams, it really is a must attend.
One of the big observations I have from the show overall is that we as an industry are still trying to figure out the mobile market and how to provide Enterprise solutions in that space. What I saw was a lot of sessions on “how to” with iOS or Android or Windows Mobile, which were well attended and necessary. But I am starting to suspect that the attendance in each of these was largely along personal favorites (iOS, Android, Windows Mobile) and perhaps not so much on need. I was surprised at the low attendance at the couple of sessions about how to pick which one is appropriate or if you want to do multiple platform deployments, how do you decide. Esri actually did a real nice job in these sessions.
Trying to put that observation or perception in context of the Keynote speakers point that the mobile market is incredibly fragmented right now, and the end users are demanding solutions working on multiple platforms is where I am struggling. Are we as technologists still so caught up in the coolness factors of each of the individual platforms and the Marketing hype that drives separation (mine is better than yours) that we aren’t seeing the bigger picture? Or is it that we are still in the early part of the technology curve where we have to understand each of the technologies first then figure out how to tie them together? That is not to say that there aren’t tools that allow us to start doing just that; it just seems like these are just getting out of the gate.
During the week before the Dev Summit I must have received 2 emails a day from some vendor, integrator, or techno think tank group offering to help me as a CTO figure out what my corporate mobile strategy was going to be. This is obviously a big emerging problem for corporate IT departments that we need to be smart about. Do they standardize on a single platform (a strategy our friend from Gartner would seem to indicate is doomed), do they allow, and therefore have to support, multiple platforms across their IT infrastructure, or do they create some sort of Enterprise Mobile Service BUS that abstracts the layer between core IT systems and whatever mobile platform you want?
Hmmm, that last one sounds a lot like the AGS Server architecture model, and by the way the same model that the run time solution is going to follow……..
Okay slight change of topic, but still related.
So I have heard a little bit about the QT development environment in the last month or two. It’s always around developing cross platform solutions for mobile phones. It has just started to show up as one of the last bullets on Esri slides about developing multi-platform solutions in mobile. But for the first time I see that Esri has fully embraced this as part of a product line in the development of SDKs for the run time solution they will release as part of 10.1. On digging into the resource center, I am seeing a modest amount of whitepapers- like instructions and getting started material for using Qt for engine development- in 10.x. Man how do you keep up with all of it!
Qt is a cross-platform application and UI framework. Using Qt and C++, you can write web-enabled applications once and deploy them across desktop, mobile and embedded operating systems without rewriting the source code. http://qt.nokia.com/. The guys presenting the run-time special session mentioned that it has been used for large scale enterprise solutions particularly when high performance is required. Some of the real-time map refresh capability they showed during the demonstrations apparently will be available first with the Qt SDK and follow with the other 2 SDKs (JAVA and Silverlight WPF). Patrick sat in a session for developing military solutions and they showed a Qt based application with military operations, moving maps, real-time comms, etc. I was also a bit surprised at the show of hands during the run-time talk, apparently there are a bunch of C++ developers in attendance. I don’t remember seeing that represented much previously.
Well over the next couple of weeks I will be doing a lot of thinking about what we saw this week and trying to figure out what it means to the industry and what we as a company want to invest in. Already looking forward to next year and what we can do until then.
The Dev Summit came to a close today but not without its share of informative sessions. I sat through one this morning that made me think more about the comment, “ArcGIS Desktop is not meant for the web”. We keep hearing how GIS is moving to the web with the many open API’s, free templates, tools for hosting and managing data, applications, etc. I also hear Esri and others say, “Desktop will always be around and there’s no way you can do the things in desktop over the web”. Not so fast; in the last year or so we’ve seen BA Desktop become BA Server with an open API, and then evolve into BA Online.
This opened the door for Community Analyst.
Early this year we saw ArcLogistics desktop become web enabled.
Now I learn that the desktop redistricting tool we’re all familiar with has moved to the web.
Each one of these tools is arguably a full featured GIS with editing data, analyzing information, map production, and dissemination. What’s next, an online property management system? Desktop over the web…not such a pipe dream any more.
I’m sitting in the airport on my way home from the summit and it seems a good time to wrap up. It’s been a great week but it’ll be nice to get home. I’ve taken in so much stuff that it’s starting to blur together but since this will be my last post for a while, I thought I’d focus on a task that’s common to a lot of the projects going on at GISi – printing high quality maps from a web application. I mentioned this one on day 3 and promised to attend some sessions on Python to learn a little more about it since that seems to be the easiest way to go about printing to a PDF from ArcGIS 10.
Python has been part of ArcGIS since release 9 and has been successful enough that support for it has been increasing with every release until now, at 10.1, it is fully integrated into the application – intellisense, code completion, and integrated help; they’ve even added a Python console to the main interface and gave it a name – ArcPy. I’ve messed with Python before, mainly by creating tools using Model Builder and exporting them to scripts that I could modify, but I’ve always used it inside of ArcMap, which isn’t very helpful when it comes to printing from the web. One of yesterday’s sessions entitled ‘Using Geoprocessing Services in a Web Application’ showed me that there’s a lot more that can be done with it, and the follow up sessions I attended on Python make it look pretty easy. Python is a very clean, easy to read language, and while I’ll have to learn a new syntax, it looks fun.
As we should with all projects, let’s first define our scope. What I’m after is the ability to allow the user to select a template from a list; this will set the locations of the map elements on the page, as well as the size and orientation. So the title, legend, scale bar, north arrow, comments, etc will be placed where we want them to go so we can make a nice clean map that’s easy to customize if we need to adjust the template. Then all we need to do is create the elements, put them in where they need to go, and output the whole thing into a PDF for the user to print at a high resolution. Piece of cake, right? If you’re doing it from scratch, maybe not so much.
Our handler needs to get the state of our map (extents, layers, etc) and the layout selected by the user, then make a REST call to a Python Geoprocessing tool that will do the work. Of course, I haven’t actually done this part yet, but since we’re dealing with ArcGIS templates, there’s existing code we can use to generate the map and export the whole thing to a pdf. The example I saw, which should be available on ESRI’s resource page in a couple of weeks if it isn’t up already, was only a couple of pages worth of code, and compared to how I’ve done this in the past using open source libraries like ITextSharp, it looked really simple.
So, as you can see, my head is still spinning from all the possibilities I’ve seen over the last couple of days, and want to play with all the new toys all at once, but the reality is there’s only so many hours in the day. Hopefully I’ll get a chance to take a closer look at some of these things in the days to come and will pass along what I learn. And finally, if you’re like me and have never had a chance to attend the developer summit before, start bugging your manager about it for next year; it’s a good time. Traveling is a pain and there’s a lot of sitting still involved (which is tough for me), but you’ll see a lot of great stuff and Esri puts on a great show.
I attended a session today geared toward the defense and intel community. The majority of the presentation centered around creating maps for military use and 2525C symbolization, however, there was a cool demo that utilized the upcoming 10.1 runtimes, specifically the C++ Qt SDK. The demo included messaging between multiple simulated vehicles. The performance of the application running on the new runtime was quite impressive.
And speaking of the new runtime, I also attended a session highlighting their upcoming release in 10.1. There are some pretty significant changes coming that will yield a smaller overall footprint, increase performance, and simplify deployment. However, low level access to ArcObjects will not be supported in the runtime. So there will be a shift in development paradigms mainly toward WPF. A word of advice I picked up in the presentation is to get familiar with the Silverlight Web API, because if you are comfortable with that, you’ll have a good head start on working with the new runtimes.
I started the final day of the conference with the ‘Building RIA Clients Using ArcGIS Workflow Manager REST Services.’ Esri is calling AGS Server Workflow Manager the ‘complete GIS management system’ and recommends using the product to create and manage GIS project workflows. The product can be used by all stake holders including clients, managers, analysts, and developers. Workflow Manager will allow you to author and publish project workflows, centralize project information and documentation, control access to data (including spatial data in your SDE repository), and integrate GIS into business processes. The main intent of this session was to demonstrate how you can build a custom workflow manager solution using the new REST endpoints available. The presenters showed examples of a custom Flex viewer that was able to query users, jobs, reports, and workflows through attribute criteria or by spatial area-of-interest.
My next session of the day was titled ‘Redesigning Desktop Applications for the Web.’ The session was basically a talk on why you would (and wouldn’t) want to move your applications to the web. The web has become a pervasive, accessible, and proven platform and now might be the best time to consider moving desktop apps over, but, ultimately, you will need good project planning and a sound understanding of the user audience to make the best decision. Some of the key information I took away from this session included:
- The web is a fast and cheap infrastructure that provides a platform for easy collaboration and sharing.
- Application development is easier now than ever before (with WebAPI’s and Application Builder tools).
- The web interfaces are richer and smoother, allowing a desktop-like experience.
- The user experience on the web is highly important. You need a simple, but powerful application.
- You must know the user audience and application requirements. In some situations, a desktop application will still be the best choice.
- Must consider security implications on the web.
- Planning is key! Planning the design, development, and deployment phases are critical to any application’s success.
My last session of the conference was ‘Advanced Development with ArcGIS API for MS Silverlight/WPF’. Having no development experience with Silverlight, my goal was to learn what is possible with the framework and how it may differ from the other WebAPI’s. The presenters here, like in most all other web sessions, stressed the importance of the user experience and reminded the audience that the application does not always have to be map-centric. The application examples presented here demonstrated the grid/map interaction (i.e. hover over a data record and an assortment of animated effects highlight the related map feature) and other visually rich ways of interacting with the map and attribute data. Some basic editing examples were also shown. From the standpoint of a beginner Silverlight developer, I was most interested in the new Silverlight Application Builder (a WYSIWYG Silverlight editor) and the Visual Studio Silverlight templates. Apparently, you can easily switch back and forth between the Application Builder and your custom Visual Studio-based Silverlight projects, which could make development in teams easier.
This was the end of the conference for me. Esri did a great job and presented a lot of useful information regarding their products and future trends.