Doron Yaacoby writes that he’s absolutely frustrated with ESRI’s Web ADF.
As I’ve said before, this is one horrific development platform. I’ve started to seriously think about coding a replacement that will work against ArcObjects (at least something that will satisfy my team’s needs), but it would take a lot of time which we currently do not posses. One day, though, one day.
I think most of us have felt this way. He had earlier said this about the ADF:
…the API is riddled with bad design choices and bugs. Weird exceptions get thrown. You find out that there are 5 different classes that are called Converter in this API, in 5 different namespaces, which in some situations can definitely be used together. Documentation for most methods is something like: Map.GetFunctionalities - “Returns the functionalities of the map.” Oh really? Now I couldn’t guess that on my own. I should mention that since the initial release the documentation has improved a lot, but it is still no paradise. Recent documentation for this method (added since Service Pack 2, if I am not mistaken), explains that it returns a collection of IMapFunctionality. So why, why oh why, does this method has a return value of IEnumerable? This is a .net 2.0 only library, for pit’s sake, they can use generics. And they do, in many other places. Here, they don’t. This kind of API inconsistency can drive a man insane, I tell you.
The WebADF which was supposed to help ArcGIS Server developers ends up being more of a hindrance to them. We looked at the ADF early on for our products but my lead developer threw up his hands and said no more and I spent a whole week trying to write a WebTask and it was virtually impossible to figure out. Sure it works great for ESRI’s example but when you try and extend beyond what they’ve documented you might as well be coding with your eyes closed.
I haven’t tried coding with the WebADF with a bucket on my head yet
This brings up a bigger issue with ESRI products lately. With long lead times between versions, we get so much more added functionality in each release. So rather than small jumps where feedback can help grow the product we are left with frameworks being dumped on us and we have to figure it out because of poor documentation. Sure 9.3 is supposed to fix all this, but 9.2 has been out since November 2006 and if 9.3 is released to beta right before the Dev Summit how much do you want to bet that the final comes out near November 2008, almost 2 years later. So in the mean time, developers such as me get so damn frustrated working with ArcGIS Server WebADF that we toss it aside. And don’t get me started on ArcGIS Server 64bit.
As Doron says:
For the amount of money an ArcGIS Server license costs, a .NET developer like me could very well expect a decent .NET API to work with. The Web ADF is a .NET API, but it is hardly as powerful as ArcObjects are, nor is it any kind of decent.
Could all this be why so many ArcGIS Server developers I know are looking at SharpMap?
90 responses so far ↓
1 Lefty // Jan 30, 2008 at 12:49 pm
Holy shit James, you sound kinda pissed.
I half expected to read, “When I get to the DevSummit I’m going to punch Art Haddad right in the nose”.
2 James Fee // Jan 30, 2008 at 12:51 pm
I can’t punch Art in the nose, he’s got a .NET SIG to run.
3 Dave Bouwman // Jan 30, 2008 at 12:58 pm
Ah! Clearly our problem is that we’ve been using cardboard boxes - not buckets! Off to Ace Hardware to pick some up!
This is a great post, and pretty much sums up how our team is feeling. “Coding with your eyes closed” is a very accurate description.
I feel for the ESRI dev teams who put a lot of energy into building the ADF, but unfortunately it is just not cutting it. And while there is a ramp up time, and it’s not exactly for noobs, when you have a ground swell of knowledgeable developers saying that something is broken, they are likely right.
Maybe at 9.3 they should just can the ADF web controls etc, and focus on exposing services that can be consumed in other clients - OpenLayers, Virtual Earth, Google Maps/Earth etc. It would make our lives much easier.
Dave
4 Ron // Jan 30, 2008 at 1:01 pm
I’ve been working with ArcGIS Image Server now for the past few weeks and I’m having the same frustrations. It’s buggy, poorly documented, takes a lot of workarounds, etc. On the bright side it is fast as hell once you get your images added to it.
5 Grabner // Jan 30, 2008 at 1:04 pm
We’ve been going round and round with ESRI on why the Editing task won’t work with there fantasy ArcGIS Server Web Application. ArcGIS Server and the ADF web controls are starting to turn out to be a black hole of problems.
Greg
6 Rick // Jan 30, 2008 at 1:15 pm
I hate to pile on because you’ve said basically what I’m thinking. Why is the WebADF so difficult? I thought the future was supposed to make coding easier, not harder.
We had a client come to us asking if we could do anything with ArcGIS Server Java ADF. I got a copy of EDN for our Java guy and he came in my office a couple hours later with a look on his face like “Are you kidding me?”.
7 Danny J // Jan 30, 2008 at 1:17 pm
See this is what really gets me angry about ESRI. They say they want feedback. Everyone I know didn’t ask for a heavy WebADF. We wanted something light and easy to use.
I’m saying this loud and clear, ESRI only listens to their large corporate/government clients and pretends to listen to average users. If they listed to the people in the trenches on this stuff, they’d have a good product.
8 KipterUh // Jan 30, 2008 at 1:22 pm
DAMN! What a post. I haven’t really thought about why I’m not using the WebADF, but you pretty much hit the big points. It is so difficult to use. I know that no one likes to write documentation, but they have to realize that those who want to use the WebADF need something to learn from or at a minimum figure out what the stuff does.
9 Sean Gillies // Jan 30, 2008 at 1:25 pm
Oh, snap!
10 Arnold // Jan 30, 2008 at 1:42 pm
Heck you can’t even search their documenation anyway.
11 Matthew S // Jan 30, 2008 at 1:47 pm
Are you really surprised that webadf stinks so badly? ArcGIS has been unstable and over complicated for a long long time. How was adding even more layers to the ArcGIS stack supposed to make it any better? Even if the api was good it wouldn’t suddenly become any more usable. Stuffing so much functionality into a single package is foolish. There is just to much that can go wrong.
12 Dave Smith // Jan 30, 2008 at 1:51 pm
The biggest issue is that there’s a steep curve from one release to the next, and it feels like reinventing the wheel each time.
I’m with Dave Bouwman on this - focus on open APIs that allow the services and functionality be accessed from the outside.
13 anon // Jan 30, 2008 at 1:54 pm
This is all h proof you need to know how crappy WebADF sites are:
http://gis.rtd-denver.com/MapServer/
I mean how do I route?
14 Casey // Jan 30, 2008 at 1:55 pm
I’m currently still annoyed that I cant make ArcExplorer read directly from SDE.
Oh, and I wish ESRI would listen to the government…or rather the government would listen to the actual GIS people.
15 anon // Jan 30, 2008 at 2:43 pm
Quite the turd James.
http://twitter.com/cageyjames/statuses/660448682
16 Wright, David E. // Jan 30, 2008 at 4:39 pm
We have spent the past year and change making things work. We now have a fluid path to build and expand on the ADF. But it cost a lot of time and money to have a solution that works.
The key was getting away from the task framework and building your own engine to work against!
Thoughts?
17 Brendan Hemens // Jan 30, 2008 at 4:53 pm
A couple of my coworkers attended an ESRI ArcObjects course recently, where they were told ESRI’s moving to VB.NET at 10, and they’re completely reworking ArcGIS Server from the ground up. Maybe this is old news, but we wonder why we would continue to pour development effort into Server if the knowledge we acquire isn’t going to be worth anything shortly anyway (the 2-year cycle comment notwithstanding). We’re simple folk, and our clients wanted relatively simple things - we trashed months of Server development effort and got where we wanted in about 2 weeks with mapserver and ka-map.
18 Chris // Jan 30, 2008 at 5:13 pm
I agree that it has been a very frustrating experience so far. From the lack of decent documentation, to the service packs that don’t fix the issues they claim to. It seems like the whole 9.2 experience has been one big beta test for 9.3 - except that beta products don’t generally come with such a whopping price tag!
I guess the moral of the story is to always wait for odd-number ESRI releases.
I feel used and abused…
19 beepbob // Jan 30, 2008 at 5:44 pm
A couple of my coworkers attended an ESRI ArcObjects course recently, where they were told ESRI’s moving to VB.NET at 10, and they’re completely reworking ArcGIS Server from the ground up.
personally, I’ve had it with this crap! All I ever hear is how ESRI is going to fix something in the future. Then, when the future comes, they say how they are going to fix the thing that is broken in the present.
They are a classic case of the carrot and the donkey. We’ll never get a stable product, but will continue to be raped by these guys. They are obviously in over their heads, and that is why they keep stringing us along.
I wish the Open Source guys had an infusion of ca$h so they could speed up development, and we could relieve ourselves of ESRI.
I swear they have more marketers than programmers…
..I’m done!
20 Tim Maddle // Jan 30, 2008 at 5:55 pm
I am desperately trying to keep my organization from shoving the Web ADF down my throat. It’s a bloated monstronsity and I suspect most people could develop a site from scratch in less time than it takes to tame the Web ADF. In addition, when something goes wrong, I bet it would be 100x easier to fix a problem.
The strange thing is that the WebADF that I saw a preview of a few months before the 9.2 release at least looked cool. The current Web ADF looks like it will be the source of the next generation of fugly map sites.
Is it any wonder that ESRI has to offer a $15000 prize for its code challenge? They’re probably desperate for someone to send in some code that shows the Web ADF is more than a pile of steaming you-know-what.
The only ESRI product that really interests me is the Image Server, and that’s probably because I haven’t actually used it.
21 CuriousGeorge // Jan 30, 2008 at 6:19 pm
One of the biggest issues I see from this post is in extending the ADF with Tasks and then there is the general ESRI bashing. What I gleam from the docs (which has gotten much better thru the sp’s) is that you really need to understand ASP.NET development to extend the ADF; including the task framework. Although I will admit, the Task Framework does need to be greatly simplified - not many people like writing Composite controls.
Everything I have seen shows that the ADF is built on top of ASP.NET and assumes if you are working with it, that you understand the Microsoft web technologies (i.e. ASP.NET and Control development) to do anything advanced.
I read in the comments that people want to code directly against ArcObjects. Well, DCOM SUCKS and should never be used in a web app directly, use the services (soap, roll your own REST) directly and cache the hell out of things for performance reasons. For some reason, GISers tend to want the web to be more than it really is.
22 David Stevenson // Jan 30, 2008 at 6:22 pm
We’ve also experienced some frustration with Web ADF over the last twelve months while building a product that integrates with it, but we have no regrets about choosing a Web ADF-based direction. I, for one, am happy to advocate that our customers leverage the immense amount of work that has gone into ADF, fix some of its problems where we can, and add missing functionality where it makes sense. It’s far from perfect, but rest assured that it’s only going to get better, and as it does, we’ll be able to integrate in ESRI’s new functionality in a fraction of the time that it would take to recreate it.
23 Lefty // Jan 30, 2008 at 6:44 pm
@CuriousGeorge: Bashing? I see some pretty good examples of why the WebADF is bad for GIS web developers. The only bashing I see is from you to us folks who don’t like the WebADF and how we don’t understand Microsoft web technologies. I can’t speak for others but I’m pretty sure I know as much as you do.
@David Stevenson: You do realize that 9.3 is not backwards compatible with 9.2 so you’ll have to recode everything. Seems like 9.2 is a complete waste of time.
24 Curious George // Jan 30, 2008 at 7:18 pm
@Lefty: My comment was not intended to bash anyone. However, if you have problem with people bashing, then perhaps you should not throw stones yourself.
As for 9.3, I got it from high authority that it WILL work - there is even handy migration utilities and docs for you in the cases that adding an additional assembly won’t work straight out the box. So, Lefty, get your facts straight before you spread ugly rumors.
25 Lefty // Jan 30, 2008 at 7:21 pm
Migration isn’t the same as compatible. I have it directly from my local ESRI rep that 9.3 WebADF won’t work with 9.2 servers and 9.2 won’t work with 9.3. Unless you work on the ArcGIS Server team then you don’t have any more clue about it than I do and I trust my ESRI rep who I’ve worked with probably longer than most people reading this blog have heard about GIS.
26 James Fee // Jan 30, 2008 at 7:26 pm
Guys, lets not get all bent out of shape here. ESRI hasn’t “officially” said if the ADF will be backwards compatible or not so lets not go bragging about who knows who because you both are spreading rumors.
We’ll see what happens at the DevSummit and what plans ESRI has for the WebADF. Don’t make me be the voice of reason here because I’m sure not that kind of a person.
27 Mellisa Maples // Jan 30, 2008 at 7:27 pm
Are you going to the Dev Summit James? I hope to see you.
28 James Fee // Jan 30, 2008 at 7:30 pm
I know I’m going but I’m waiting to see how many in our company can come before spending the money on registration. I think I still have some time before the early bird registration ends.
29 David Stevenson // Jan 30, 2008 at 8:04 pm
@Lefty. Although ESRI might have made some changes since we did our holistic testing in Redlands a couple months back, it took us about a day to migrate about ten thousand hours of development work done with 9.2 an early 9.3 build. As James pointed out, there’s no official word yet, but compatibility concerns aren’t keeping me up at night.
30 Steve Larch // Jan 30, 2008 at 9:02 pm
Well shoot… my company is just now making the transition to 9.2 so I haven’t had a chance to play with the ADF. I keep thinking about begging, or paying out of pocket, for an EDN subscription just so I can stay up-to-date but this really makes that $1500+ dollar pill hard to swallow (or explain to the wife).
So what is the most common ArcGIS Server/ADF alternative? MapServer/Openlayers? SharpMap is new to me but I’ll be checking it out over the next few days.
Thanks for the info James.
31 Mike // Jan 30, 2008 at 9:29 pm
[Danny J wrote: "I’m saying this loud and clear, ESRI only listens to their large corporate/government clients and pretends to listen to average users. If they listed to the people in the trenches on this stuff, they’d have a good product."]
Danny, I think you overestimate how much sway large customers really have with ESRI…
32 anon // Jan 30, 2008 at 9:49 pm
I think you overestimate how much sway government customers have with ESRI as well…
33 Reid Watkins // Jan 30, 2008 at 10:16 pm
Regarding compatibility between 9.2 and 9.3, James provided a link last week ( http://www.esri.com/news/podcasts/audio/speaker/devsummit08-haddad.mp3 ) to a podcast where Art Haddad and Rex Hanson discuss their upcoming .NET ADF presentation at the Developer Summit. About 3/4 of the way through Rex says, “Things that you built at 9.2 should just work at 9.3.” Does that count as official?
34 Brian Johnson // Jan 30, 2008 at 10:40 pm
So…coming at this discussion from a different direction. I can bash ESRI with the best of them, and the ‘help’ I’ve received from ESRI support is enough to make me cry.
Looking at my driver’s license, I can determine that I am older than I feel - but in the programming world I feel I am slowing down, and in the ESRI world I just don’t get it.
I’ve been a hobby programmer since Integer Basic on the Apple ][. I been around for Pascal and FORTRAN (but I missed out on C or C++). I learned VBA by following recorded Excel macros, modifying them to see what happens, and learning most of the Excel object model throught the help file - of which, v97 was the best. I got Visual Studio 2005 2 years ago so I could help out on some programming work within our company. But - my company’s focus is not programming, but environmental consulting. So - my ability to learn on the job is very limited when I need to keep that utilization rate up.
Once I became involved in the GIS world at ArcGIS 8.1 - I so wanted to automate some tasks in VBA. Things that were repetitive - and I knew there had to be a better way. But - my old standby for learning the object model, the macro recorder, doesn’t exist in the ESRI world. So - I tried the help files - not much help there. I tried downloading ArcScripts done in VBA and tearing them apart - but I just couldn’t get it. The 5 day ArcObjects class always seemed to be scheduled at the wrong time (and wrong price) for me to be able to attend.
Now - I was blessed with and EDN subscription last November, and since my work schedule is still full (and my kids still love to play UNO and Legos with daddy) - I still am having trouble ‘getting it’ when it comes to trying to program ESRI software of any kind.
My solutions so far have been to get the maps/figures I need for our projects done in AutoCAD. The object model there is so easy to understand, and the sample code in the ACAD VBA help files usually solve 90% of my problems. Using an Excel worksheet of my data and VBA - and I can create what I need in AutoCAD in a couple hours. Sure - AutoCAD is not GIS - but for most of my needs, it works.
Does anyone have any helpful suggestions for wrapping my brain around the ArcGIS object model? Or the Web ADF? I’ve got $30K of this EDN software to play with for a year - and I’m afraid I won’t be able to get any of my ideas out of the gate. Or even ‘Hello World’ at this point. Not much chance of getting the subscription renewed if I can’t show any progress in developing some applications our company needs.
Any ESRI documentation that is helpful to you? Any place a beginner can start to learn how to program for ESRI software? Even the user forums at ESRI sometimes go beyond my understanding.
35 Jack // Jan 30, 2008 at 11:23 pm
Those of you bashing esri need to take a long hard look at the adf documentation before making outrageous claims that the adf does not function as advertised. The folks at esri have spent millions of dollars publicizing this product and they deserve better than to be torn apart by beedy-eyed dungeons and dragons, 12-sided die throwers.
Your boss’s boss spent hours reading the marketing documentation on this product and it does work. You just need to learn the proper way to use it.
Quit your whining, come together as a community, share your ideas in an open forum, submit suggestions for improvement, and wait patiently.
The lowdown I’m getting from my sales rep is that arcgis 9.3 sp7 is going to revolutionize the gis industry. Don’t you want to be onboard when that happens?
Are you so selfish that you cannot wait for ItopologicalOperator25 to intersect your polygon without an HResult?
Ask yourself - Are your clients so desperate that they can’t click through an unhandled server exception?
Come on people!!! Now is the time to come together, join forces, rise up and take action. Wait on the next release and service packs and it will be worth your while!
The innernets web mapping concept is only just born. Can your google/live map do an intersect? can it geocode? can it store customized features?
can it lock a feature class?
To the adf naysayers I respond - You don’t know Jack!!!!!
36 Rachel // Jan 31, 2008 at 4:14 am
The move to Ajax has been tough for everyone, in the main stream dev community there’s been alot of blow back from .NET devs who now have to lean javascript and AJAX technology.
I had a big learning curve with the ADF, but I’m relatively happy with it. BUT I had to do the following things:
1) Buy O’Reilly’s Javascript bible and master JS
2) Take a hard cold look at the underlying technology running the ADF which includes a firm understanding of:
ASP.NET 2.0 Server control Development using Embedded Resources and Javascript (I actually bought a 500 lb book on Server Control Dev at 2.0)
ASP.NET 2.0 concepts like Themes and Skin Files
Understand CSS very well
Understand Ajax technology at a very fine grained level
Understand how to debug embedded resources and use Visual Studio tools like script explorer.
Once you get these concepts, the ADF really aint all that bad. And I’m sure people will say well you shouldn’t have to learn all this stuff, but yeah I do, I’m a developer, I have to keep ontop of all this. If i didn’t touch esri technology, as an ASP.NET developer I’d still be expected to know all of the above technology.
After dealing with this, the team i’m on extended the ADF very nicely (and within the task framework). We have several happy customers with the extended ADF up and running.
If you think this is bad, wait till you start developing in 3.0 and doing Silverlight, WPF and XAML. It’s a whole new bucket of stuff to learn, and it’s not easy, but I cant ignore it and hope it will go away.
37 bender // Jan 31, 2008 at 4:34 am
you can choose to use your hands to write a converter in the SharpMap project for the 5th time or use them to or virtually (!) punch some noses. i prefer the last way. it’s less work and more fun.
we are a strong community here, i think. many people are reading this blog. get a meeting at the dev summit and ask the really hard questions. collect the bugs, collect the inconsistencies in a blog entry. present them at the summit. threat esri that you won’t use the adf any more, if they don’t react on you. take it seriously. you will need an organizer and speaker. get a tie. polish your shoes. comb your hair. name yourself “the spatially adjusted community enhancing the esri technology to reach the f#!$ing project goals”. get a logo. get a shorter name for this.
just a bit kidding
root planting is not the way it goes here, man!
38 Johnny R // Jan 31, 2008 at 6:55 am
Heavy sigh… I remember when ArcObjects was just as vacuous, poorly documented, buggy and stubbed. It was murder coding for events in ArcInfo 8 - and to watch a bunch of AML and Avenue hacks take a crack at COM - FUGGETABOUTIT!
ESRI products - much like the works of their evil step parents up in Redmond - are works in progress. Now - ArcObjects is probably the most fine grained spatial data API framework in the friggin World. I remember when AI8 came out, there were rumors floating around that there were more lines of C in AO than in Windows NT. But I digress…
All of this stuff can be encapsulated and re-engineered with the right IDE and language - pick your poison. I’ve heard people bitch for years about the move to COM, to AJAX, the freakin JavaScript spiderweb of the ArcIMS HTML Viewer, the clunky Java Viewer (long may it rest) - the list goes on.
Hey man, look at it this way - you used to have to fork out like $15K for a Sparc to run this crap. Now it sings on a Dell lappie - I mean, COME ON! SanDiego and Palm Springs!!!!!! ‘Nuff Said.
JR
39 Marianne // Jan 31, 2008 at 6:55 am
I don’t know about all of you, but going from Avenue to ArcObjects & VB6 a few years ago wasn’t an easy feat. I remember it taking me hours to figure out how to create a tool in VB6 to add to a toolbar that would then be added to ArcMap or ArcCatalog. Of course, I have Geography degree, not a CS degree, so maybe that has something to do with it… Over time, though, I became very comfortable with ArcObjects and learned to love it (it’s easy when you know it well).
Moving from the Desktop world to the Web world has been similarly difficult (although even harder). It’s a whole different way of programming. You don’t need to know just one language, you need to know at least 3. Then there’s the Web ADF, which was completely transformed from 9.1 to 9.2, and very poorly documented. I’m almost afraid of plunging myself too much into it from fear of having to learn it all over again when 9.3 or 10 gets released. ArcObjects has pretty much not changed since the early days of version 8 (and is probably why developers want to use it with Server as we know it well). But then, the Desktop environment hasn’t either. The Internet, however, is not the same it was 7 years ago. People expect web applications to duplicate Desktop functionality, which is just not realistic at this point.
I curse ESRI on a regular basis, don’t get me wrong, but the reality is that their software can do a LOT. Yeah, there’s OpenLayers, MapDotNet, Google, Virtual Earth, but they don’t have the kind of power (yet) that ArcGIS Server has. I think each platform has its strengths and weaknesses. We, as developers, have to learn those and figure out which one makes the most sense for the application we’re developing. But that’s all very overwhelming. We all have deadlines, we have budgets, we have to be x% billable, etc. And honestly, I don’t want to spend my nights and weekends learning this stuff when I can spend that time with my kids.
I think the problem here is that ESRI is not making it easy for us to learn the WebADF. I thought ArcObjects was relatively well documented. It was no ArcInfo workstation documentation, but it was alright. ArcGIS Server 9.2, however, was very poorly documented. How many classes were completely devoid of any explanation? Probably the vast majority. The only way of learning what a class did was to see an example of it being used. I know, I know, it’s gotten better with each SP, at least so people say. I still haven’t noticed much difference. I would rather have ESRI delay the release of a new version by a couple of months so they can document their product better than to rush it to us in its current state. They also need to provide better training. Maybe even combine training with Microsoft? For those of us moving from the Desktop/ArcObjects world to the web/WebADF world, it’s quite a jump (at least it was for me). Provide us with some good foundation, or tell us which technologies we need to be familiar with, and possibly also recommend training from 3rd party providers. It might not be their place to do this, but it will help their cause if they help us out.
They also need to provide better support for programmers after a big release. I was able to call them for support with ArcGIS Server, but it would often take 1 month before they got back to me. My deadline had long passed by then. And I do realize it’s hard to find qualified people that know this stuff.
Finally, I think that ESRI needs to implement some internal standards. Force programmers to document their code as they go. Then you won’t get stuck at the end with lots of classes (including many “Converter” classes) that are not described. It’s very simple really.
40 CuriousGeorge // Jan 31, 2008 at 6:57 am
@Rachel: Well said. Developers really need to understand MS Web Technologies and then the Web ADF will make more sense.
@James: I like benders ideas. It seems you know folks on the ADF team, can you contact them and ask them to get a gathering to discuss these things at the dev summit. I am sure Art would be happy to do this - maybe at his .NET SIG or after one of his an Rex’s sessions on the ADF???
41 Jason Harris // Jan 31, 2008 at 7:16 am
My issue with 9.2 and earlier versions is that the vast majority of web-centric developers are left out with no way to hook in. What are php developers supposed to do? Coldfusion users can theoretically use the Java api, but you all of you are complaining about the .net docs…you should look at the java docs - absolutely terrible. If you are interested, you can read some of my rants and comments over at http://www.roktech.net/devblog/index.cfm?mode=cat&catid=8A18B996-105A-1791-BADB8BA9A65944F0
The point is, your life becomes miserable if you leave that adf. Personally, as a developer, I need to distinguish my apps from others - and I’m not going to be able to achieve that level using a drag and drop ADF.
My sincere hope is that ESRI gets its act together with 9.3 via the rest & js api. If not, I’m afraid there with be a lot more googlemap developers on the market soon…
42 Jack // Jan 31, 2008 at 7:29 am
New releases are great. Service packs are great when they fix things. Adding new functionality into service packs is just bad form. Back compatiblity is golden.
Any developer who has had to maintain a dev environment for their esri clients will tell you that more frequent service packs is a nightmare. When 9.3 comes out we will need a whole new series of machines to support each client configuration.
43 Schuller Tom // Jan 31, 2008 at 7:47 am
I’m a pure Java developer and have just started with WebADF. So I can’t give any positive or negatice remarks on it.
But I was surprised to be forced by Esri to use JavaScript to develop a Java GIS webapplication! A GIS webApplication which is promoted by Esri to be build up by JSF-components!
So I started on my own an Icefaces-Esri-Integration-project. Icefaces is a JSF-library which handels all the Javascript code inside the JSF-tags. To customize or extend a component, you only have to develop it in Java.
I have actually rewritten these tags:
‘map’, ‘toc’, ‘tool’ and ‘command’
Icefaces uses D2D (Direct2Dom) which never reloads the complete page but only the modified parts.
You can visit my demo application at:
http://www2.pch.etat.lu/pchViewer
Everything is build with clean Java and JSF code. No Javascript coding is needed to build that application!
What do you JavaGuys think?
Any feedbacks or interests are welcome.
Tom
(see you at PalmSprings)
44 Jason Harris // Jan 31, 2008 at 8:12 am
@Tom:
java.lang.OutOfMemoryError: Java heap space
45 Janene // Jan 31, 2008 at 9:24 am
I tried to use the WebADF last year to put a simple map on one of our web applications. I failed - could not get the functionality working well enough in the generous time I was given. I have not failed this badly in 20+ years of being a software engineer. It shouldn’t be this hard! Our app still doesn’t have a map on it as we are non-profit and ran out of funding to work on it for now.
See you all at Palm Springs!
46 jxn // Jan 31, 2008 at 9:53 am
I you know asp.net, javascript, vb.net, and learn the client callback framework, you can work thru any problems you have with the ADF. It will take 4 times as long as you thought it would.
Just dont ask for help or documentation
47 Mark // Jan 31, 2008 at 10:58 am
After working with ESRI for the past 15 years, I’ve come to know their idiosynchracies. I think the Web ADF will eventually become more stable and well documented in a year or two. I’ve deployed a few sites, and not without pain, but the user’s forums usually save me from insanity. For now, it’s a bit of a guessing game with that ArcGIS server product, and those folks who don’t have the time or patience because they are in the private sector, and need to make actual cash, should probably consider their alternatives, and they’re out there. I broke in to SharpMap yesterday, and find it an extremely interesting alternative, in some capacities, for the private sector work I do on the sly. I am very happy SharpMap exists.
48 Brendan Hemens // Jan 31, 2008 at 7:22 pm
The thread (though very productive) is pretty much dead, but I couldn’t pass up the chance at a further comment.
I think free and open source is like school - I did well because I worked hard, but also because I shared everything I did (a common experience in ‘applied’ faculties, I’ve found). In return, I got ideas and solutions from others I never would have thought of on my own. ESRI’s pay-for-everything model that only exposes development to a small number of people greatly restricts innovation. We can play the ‘game’, and learn what we’re told to learn, or we can learn what we WANT to learn, and share!
There seems to be an eternal dalliance with open source in the GIS community (and in this blog - James makes constant references to open source projects - for which I’m thankful - which I might never have discovered on my own). Do the brave thing - instead of spending money to figure out how to use something you had to pay for, spend money to figure out how to use what’s free. In return, benefit from the experience of others, instead of having to pay consultants to figure it out for you.
I recognize that many of you are working in a very different environment than me (provincial/state government), so it can’t apply for everyone - but how many of you continually keep one eye on alternative solutions, tracking them as bizarre proofs of your continual frustrations with ESRI, yet never making the jump?
I think I’m ready to jump!
49 Anonymous // Jan 31, 2008 at 11:40 pm
Interesting that no one has mentioned the ironic fact that the original developer of SharpMap, Morten Nielsen, is now a developer on the .NET Web ADF team. Hopefully this bodes well for the 9.3 release!
50 Andy // Feb 1, 2008 at 7:15 am
Couldn’t agree more. You think the .NET WebADF is bad try the Java one this is even worse.
I think it is a poor show especially as in the UK for a distributed ArcGIS Server install you have to pay as much for the WebADF as you do for the SOM & SOC components. I would have expected a much more robust and well documented product.
Maybe we should all just use a Google Earth front-end or similar.
51 MapsRus // Feb 1, 2008 at 8:04 am
I couldn’t help but comment on this thread as well. Is it just me, or, has every demo site, or live implementation of ArcGIS Server had performance issues? Even when using tiles, the performance does not compare with Google or Yahoo. I think I’ll stick with ArcIMS for a little longer. Perhaps the dual threaded, 64 bit environment ArcGIS 10 is promising will be worth the wait?
52 Dave Smith // Feb 1, 2008 at 8:21 am
@MapsRus: in the case of Google, Yahoo, Virtual Earth, OpenLayers what you essentially have serving up the base mapping is NOT a mapping engine, but instead a cache optimized for serving up prerendered images. Essentially, it reaches into RDBMS, file-based storage or similar repository and returns the image tiles. That’s functionality that doesn’t need to and perhaps shouldn’t be residing in the same stack and pipeline with the map rendering, geoprocessing and other more demanding processes.
53 James Fee // Feb 1, 2008 at 8:22 am
I can safely say ArcGIS Server outperforms ArcIMS and can be as fast as GM or VE. The problem is that demos are usually not run on the best hardware. I’m pretty sure the AGS development team would love better hardware to showcase their product.
When I have AGS map cache and overlay it with VE I’ve had no speed issues whatsoever.
54 MidNight Mapper aka eil Havermale // Feb 1, 2008 at 8:43 am
All of this very acidic debate has me quite concerned as apparently ESRI has a created yet again a technology shift with poor implementation for those who’s business ecology require these products. Gettting trapped into a product test that is externalized on the developer community seems a weak commitment? But I have to admit that over the years of following ESRI development treads, this has not changed over the many, many years? And that may be the real issue?
55 Bill Thorp // Feb 1, 2008 at 10:01 am
“I have AGS map cache” … if you’re just using it for maps, you paid a lot of money for a grey-box solution that does half what BSD-licensed TileCache does.
On the geo-processing side, I agree with Curious George, DCOM should be considered harmful. One hears plenty of stories of people questionably using ArcGIS server code not in SOC, but on the webserver itself for performance reasons. Unless you REALLY need heavyweight/long-running code or ArcObjects, you don’t need AGS.
56 Bill Thorp // Feb 1, 2008 at 10:04 am
Grr… double post. Need more coffee.
57 anon // Feb 1, 2008 at 2:50 pm
Perhaps ESRI should enter the plumbing business…
Here’s a breakdown of how a local government flushed your hard earned tax dollars on failed ArcGIS Server online editing projects:
$30,000 - License
$22,500 - Maintenance for ~3 years.
$10, 000 - Dedicated Server
$5,000 - Developer Training
$80,000 - Consultant fees (for 3 failed projects)
Grand Total Expenses: $147,500
Return on Investment: Zilch, nada
58 James Fee // Feb 1, 2008 at 2:55 pm
@anon: I had a client of mine call folks who sell people that kind of software GIS Carpetbaggers. They drop in, slick presentation, pay them big buck and they basically walk away leaving you with nothing. At no point do they actually take the time to see how your organization is currently using GIS and where you want to take it. Just marketing speed, cutting checks and feeling like you were taken to the cleaners.
59 MapsRus // Feb 1, 2008 at 3:13 pm
Here are a couple of ArcGIS implementations in need of some serious hardware?
http://mapserv.utah.gov/
http://maps.bexar.org/imf/sites/cc_foreclosure_beta/jsp/launch.jsp
60 Tim Maddle // Feb 1, 2008 at 9:45 pm
@MapsRUs,
Gotta love the “Blank Map? Click to Redraw” button on the Utah site. Slow as molasses. I’m going to guess it’s not cached/tiled because it’s not practical with the number of layers they are incorporating.
The maps in the second link seem to work quickly, but the documentation says that the map uses a framework based on ArcIMS, not AGS. I really think that, better than using AGS, many people would be better just using ASP.NET to create their own web services that called ArcIMS behind the scenes.
@anon and James,
I think my organization is going to bite on the AGS Advanced based on some canned demo they’ve seen of someone adding points to a map in AGS Advanced. There seems to be something in my organization related to the letters E-S-R-I that causes people to lose a healthy level of skepticism.
61 CuriousGeorge // Feb 1, 2008 at 11:48 pm
@MapsRus: You can’t judge the ArcGIS Server or ADF software based on these poorly implemented sites. Any developer/application can easily destroy even the best framework if the implementors do not know what they are doing.
Look - The ADF is complex, yes, I agree with that, and it seems to have been released before it was ready. But, it leverages existing MS web technologies and assumes the developer using it is conversant in ASP.NET and control development as described by MS - this is a positive since I do not have to relearn what Microsoft has already built. Granted, it lacked documentation, but the ArcGIS Server blog and online doc (thru the sp’s) have become much better over the last few months.
I have been developing Web stuff for the past few years and understand the complexities of the Microsoft platform… I have seen many developers that come from the desktop world totally baffled by the web and say it is the worst thing in the world. Let’s face it, the web is not a pretty place and has many variables to overcome.
I for one am withholding judgement on the ADF until I go to the Dev Summit and hear what ESRI has to say. They are a good company and I am sure they are trying to do good and stand by this product of theirs.
I can’t be the only one that sees good in the world
62 bender // Feb 2, 2008 at 11:06 am
to sum it up: there is no product on the market yet that can be compared to ags. in it’s completeness (where the adf is an important part) it is perfect for many customers and definitely worth the price.
as we’ve discussed here: the really bad thing is the way it is developed. when creating a dizzy-tizzy-fizzy product like this you need to take more care in the development and testing processes. it may be the death for ags if we all get the clue that it was created by punch drunk developers.
the real sad story behind ags is that all of it its concepts and ideas is very fine. but it was released too early - and therefore invented two/three years too late.
63 Tim Maddle // Feb 3, 2008 at 8:26 am
@curiousgeorge,
Do you have any examples of well-implemented, publicly available ADF sites, preferably using AGS as a backend?
64 EricG // Feb 4, 2008 at 6:17 pm
So… ADF is relatively replace-able with our own front-end to SOAP services… but can you avoid SDE when most of the organization’s power users use ArcInfo to access their GDBs?
Or, rather, what’s the solution that doesn’t require ME to code up a service-provider every time a user wants a new “topology rule just like with SDE”?
65 Mike Diss-Torrance // Feb 8, 2008 at 4:39 pm
I’ve been building ESRI GIS applications for nearly a decade. I agree that every out-of-the-box product ESRI has released was less than what I was hoping. Most of them did not meet the business needs for the client.
Documentation back then had things like “a widget is a type of thing-a-ma-bob”, followed by “a thing-a-ma-bob can implement a widget”. Sounds like some document practices have not changed too much. However, with the developer network, things have gotten a lot better (not perfect, just better).
I have not begun working the ADF at 9.2, so I don’t know its drawbacks (yet).
The last application I wrote was a web based ADF-like application using ArcIMS (map presentation), ArcGIS Server (Advanced GIS processing), and an ActiveX Control hosted in a web browser (client side interaction). This project took over a year to develop. Now I want to modernize it and make a .net solution.
Since we already have ArcIMS and ArcGIS Server for our other applications, it would be cost effective to continue using them for this solution.
So here’s the thing: How bad is the ADF? While I’m comfortable with working with ArcObjects, I imagine it will still take me months (or a year) to build my own ESRI-like-ADF. Just because the ADF is far short of what it should be, would it be less time/money to build work-around of bugs or to build the application from scratch again?
Mike.
66 Lakshmanan // Feb 11, 2008 at 2:22 am
Wonderful!. am from desktop background, trying to learn ADF, but ended with pain. Now I have to learn Java Script, Web Technologies like JSP/Servlets etc in order to learn ADF. The poor documentation from ESRI on server side. Why cant they provide stuff in simple terms. ESRI thinking that all the GIS developers are expert in all technologies. ESRI please consider buddy developer like me.
67 Marek Caltik // Feb 11, 2008 at 3:10 am
.NET ADF - it can be…, as we have used pure “manual” Ajax as a GUI (no Task framework) and COM dll as an ArcObjects backend. But the result was the same functionality and twice development time.
But Java .ADF is as one of my colleagues said, made by sadists for masochists. Especially task framework is a big step back.
68 Schuller Tom // Feb 11, 2008 at 3:17 am
I have just started with the Java ADF.
What I found horrible, was that I have to learn Javascipt AND Java AND WebADF AND JSF for a JSF-WebApplication.
For me, a JSF-webapplication would work with JSF-Tags and Java-Code. No Javascript coding for any functionalities like Ajax,…
That why I started my Icefaces-Esri-integration project.
http://www2.pch.etat.lu/pchViewer
See also:
http://gisprog.wordpress.com/2008/02/11/web-adf-learning-path/
69 Mike Diss-Torrance // Feb 11, 2008 at 7:14 am
I had another thought:
Beyond the ADF and all the functionality that ArcObects provides, what is the alternative (comparing functionality)?
I’m not defending the ADF, I just would like to know what other choices are there to do high level GIS processing?
Mike.
70 Dave // Feb 11, 2008 at 2:21 pm
We’ve built a SPRING Web MVC based ArcServer application with an Ajax DHTML/Javascript client. No ADF just ArcObjects and POJO. Very extensible but it was a time sink to develop… Keep in mind ESRI doesn’t always develop software based on customer needs but IMHO rather on the interests of the software development team.
http://www.mapportal.net
71 Reid Watkins // Feb 11, 2008 at 2:54 pm
@Mike:
Spatial SQL queries directly against the database from ASP.NET. That’s my preferred pattern for developing web applications right now.
72 Tim Maddle // Feb 13, 2008 at 3:39 pm
I’m with Reid. Spatial SQL will be where it’s at, with some geoprocessing functions such as buffering available directly in the SQL statement and some available as stored procedures.
73 Good affordable desktop developer for hire // Feb 17, 2008 at 8:36 pm
I noticed that the one post that suggested we all buck-up and just be good developers by learning the necessary stuff was posted at 4 am.
I haven’t logged so many all-nighters with no results since the college bar scene as I have with the web adf.
This middle-aged, semi-retired night owl with kids and solo consulting gig may just be getting outsourced.
I steered away from IMS after working with AV 3.x map cafe. It appears to be baaaack with a vengance. Are they still hiring park rangers?
74 OrganizedChaos // Feb 18, 2008 at 1:23 pm
First off I would like to say hello to all, this is my first post here. I have some strong opinions about AGS (and other topics here) and I would like to join the discussion(s).
Some of my own observations…
It seems like AGS performs fast for simple mapping functions like map display and such when using a soap based client.
For me, just as it was with the ArcIMS Html viewer - it is the UI code that is the albatross around the neck. I have been developing some custom Ajax based UIs for use with ArcIMS and with AGS and even without image tiling AGS is fast when you cut away the ungainly UI components of the web adf.
I won’t be at the developer conference (not enough duckets) but I will hopefully be attending the ESRI 08 International Conference. I hope to have something to show way before that. So, I hope to be able to post back to here with the results.
As for the posts above regarding geoprocessing, I agree. The most interesting thing to me about AGS in its current incarnation is the geoprocessing potential that it exposes at the server level. I see this as a geoprocessing tier in development of client server application architecture and development.
One of the things that excites me about the javascript/rest api is the low opportunity cost for getting something up and going for in house developers. In my own experiences, the bulk of the GIS web app clients are city/regional/state/federal entities with vastly different capabilities for developing applications. With javascript and rest, you begin to open up some of the potential of AGS to developers with smaller resources. Incidently IMHO, I think this is one of the great things about google maps and VE, with the api you can get access to the map without all the map baggage getting in the way of traditional web developer application development.
Sorry, I rambled on a bit longer than I meant to but thanks for the forum in which to discuss such interesting topics.
-OC
75 Bill // Feb 21, 2008 at 12:22 pm
I don’t know what you guy’s are complaining about… The ArcGIS Server Web ADFs are awesome. I can pan and zoom and support 1 user. That’s all anyone needs… right?
76 JimBen // Feb 26, 2008 at 9:12 am
All of this talk about creating a new web framework that works better than ADF, or AGS works OK if you buy enough hardware, or any other forms of ESRI apologizing, is just putting lipstick on a pig.
AGS is based on ArcObjects, which is still COM technology at its heart. ESRI has added thin .NET and Java wrappers around the original (poorly designed) API, but its guts are still COM. Someday, Microsoft is going to say “COM doesn’t work anymore on Windows, it has to be .NET top to bottom”. It may be years from now, but that will force a rewrite of AO (good riddence!). The API that you’ve invested a good chunk of your career to master will go into the same cardboard box contains AML, Avenue and MapObjects. If history is any indicator, that replacement for ArcObjects will be poorly thought out, poorly documented and buggy. Back to square one.
Zoom out a couple of levels and look at the entire ESRI stack. ArcObjects? Ten-year old technology. ArcSDE? Ten-year-old technology. Yeah, yeah, you say they’ve made significant improvements in each release, but what the ESRI cheerleaders never admit is that the fundamental flaws are never corrected, and the abused user base is left with unkept promises and a big bill every year. Requesting product enhancements seems more like sending a wish list to Santa Claus, rather than legitimate feedback to a vendor that doesn’t take its customers for granted.
Zoom out another couple of levels and ask, “What happens to ESRI when Jack D. retires?”. Does he have his successor picked out? Will ESRI go public? Will control be divvied up amongst Jack’s inner circle of trusted lieutenants? Will they sell out to Oracle or Microsoft? Nobody knows, but I’ll just betcha there will be a sudden change in corporate direction, and whatever you expected from the product lines will change. Think about it folks, you’re betting your careers on a organization that’s all about one guy!
My advice is to move to an OGC-compliant open-source stack. No hidden agendas, no schmoozing marketing calls from GIS Carpetbaggers, no orphaned libraries, no black-box mystery, no license management games, no more architecting the system to minimize licensing costs.
I’m not pretending that open-source isn’t work; there’s lots of effort in choosing, mastering and integrating things like WMS implementations like GeoServer, MapGuide or Mapserver, with spatially-enabled databases like PostGIS and client libraries like OpenLayers or MapBuilder
But the effort you invest in an open-source stack won’t become throwaway simply because some vendor CEO decided that he can improve his profit margin by orphaning a product line.
77 ChrisW // Feb 26, 2008 at 9:50 am
“Will they sell out to Oracle or Microsoft?”
Google!
78 Eric // Feb 26, 2008 at 10:36 am
While there are many criticisms of the Web.ADF, for those of us who have suffered these past few years through the limitations of programming with the Active X API that came with ArcIMS, the Web.ADF is a relief. It’s like the Berlin Wall falling for us .Net programmers.
The crazy proliferation of namespaces is a stregic mistake and a missed opportunity. A “polygon” should be a “polygon,” regardless if it is from an ArcIMS service, ArcGIS Server local service, or ArcGIS Server internet service, or the web browser graphic, etc …. (each is a different name space).
79 ChrisA // Feb 26, 2008 at 10:55 am
I hate to say it, but MapObjects IMS circa 1998 was probably the best web mapping product ESRI ever made. You could build whatever you wanted around it with whatever technology you wanted and I used it to serve highly detailed parcel maps to thousands of users. Simplicity worked best, but of course it didn’t have the highest license costs…
80 JimBen // Feb 26, 2008 at 11:27 am
Google’s too smart to buy ESRI.
What they might do is bluff with an inflated offer just to force M$ to overbid and spend too much.
LOL
81 Tim Maddle // Feb 26, 2008 at 5:52 pm
@JimBen
You make lots of great points in your long post, especially the last paragraph. Part of the reason why I gravitate towards Open Source is a feeling that you don’t get all the bullcrap propaganda surrounding the commercial products. No grandiose claims of online editing and geoprocessing and georeplication that turn out to be murder to actually implement. My organization (against my wishes) upgraded to 9.2 based on all the hype about the new features, only to find out we’d have to invest tens of thousands of additional dollars to actually purchase the required licensing.
@ChrisA,
I didn’t get to work with MO IMS, but I’m not surprised to read that it worked so well. ESRI really seems to be going backward in terms of performance. Sure, you can tile with AGS to improve performance, but tiling is a problem because my organization has so many layers that are not fusable and the attitude that everything must come directly from the SDE. AGS, out of the box, should give absolutely incredible performance. To the degree it doesn’t offer that performance out of the box, or that I have to work to get that performance, I don’t see the reason for sticking with AGS over choosing an Open Source (or lower priced) alternative.
82 Jamie // Feb 28, 2008 at 6:34 am
I’d like to offer another contrary opinion on the Web.ADF. I work for one of the GIS carpetbaggers but, in our defense, we’ve been in the trenches for almost a decade building working solutions on IMS and, now, Server. Three years ago we built a custom AJAX API for ArcIMS and now, after working with the ADF and ESRI’s off-the-shelf controls, we’ve done the same thing for ArcGIS server. What our team has designed is a fast, robust and stable .NET AJAX control suite, called MapsDirect, built entirely on the ADF. We developed the MapsDirect controls and solutions to fit into a niche between ESRI’s off-the-shelf ArcGIS Server controls and its forthcoming REST API.
MapsDirect includes a robust JavaScript API that moves the majority of application intelligence client-side. For instance, it renders geometry (selections, mark-ups, etc) as VML or SVG, which allows for a lot of stuff to happen in the browser without too many server postbacks.
The web site is http://www.mapsdirect.net but I encourage folks to look at our live demo application at http://demo.mapsdirect.net/. The demo runs on three AGS map services - one published ArcMap document for dynamic data and two map caches for static roads and imagery.
Our tools weren’t built in a vacuum - they were designed to fit the needs of about a dozen clients we are working with that need them and are in production. For a sample production application, take a look at http://128.172.160.131/gems2/.
83 Mike // Feb 28, 2008 at 10:56 am
Has anyone used 9.3 beta? Does it really integrate better with ASP.NET AJAX (like being able to use Web ADF controls inside of an asp:UpdatePanel)? It seems like I run into a roadblock or have to find a workaround for almost eveything I try to do.
Also, has anyone built (or better yet, know of a public web site that I can see) that has anything other than a relatively simple Web ADF app? We want to use it for an enterprise-wide GIS, but I’m beginning to doubt that it is possible without a herculean effort. When I first saw the name of this blog, I chuckled…now I’m beginning to take it serioulsy.
84 Dave Bouwman // Feb 29, 2008 at 7:34 am
@Mike,
I have not used the beta, but we are developing an open source REST connector (see http://www.arcdeveloper.net for more info) which we are using with Virtual Earth as a front end. We would be using OpenLayers but I can’t get a definitive answer of the legitimacy of using VE in OpenLayers (client has global scope and needs the base data)
As for using the current ADF for an enterprise GIS implementation (i.e. taking ArcView off the desktop and cramming it into a browser) - good luck! I’ve heard of some pretty burly focused applications that use AGS for spatial analysis and ArcIMS for display, but nothing in the wild yet.
I have high hopes and low expectations for the 9.3 .NET ADF. The Javascript client - rumored to be based on Dojo - may be a more viable and flexible platform…
85 CuriousGeorge // Feb 29, 2008 at 3:30 pm
@Dave: I have seen 9.3 Web ADF Beta - it rocks. Much improved over 9.2 and yes, integrated with ASP.NET AJAX with support for working with triggers in the UpdatePanel.
Hell, it even has a publicly documented JavaScript object model built on ASP.NET AJAX which makes working with it in the client a joy.
86 CuriousGeorge // Feb 29, 2008 at 3:37 pm
@Jamie - Your app is ok, but I think the 9.3 Web ADF is still better…. let’s get to the dev summit and see what the team presents.
87 CuriousGeorge // Mar 1, 2008 at 8:05 am
@Mike: UpdatePanels are for making non-ajax controls easier to work with - the ADF controls are already AJAX controls (the 9.3 Map control is now a Scriptable Control). There is no need to place any AJAX control inside of an update panel. Instead, you need to use triggers to allow for bi-directional communication between AJAX controls. That would be more efficient anyway as the Update panel is essentially causing a postback of the whole page that is filtered onthe way back.
88 Tim Maddle // Mar 1, 2008 at 9:37 am
@Dave,
I have the same hopes and expectations for the 9.3 ADF. Although I know it won’t happen in the near term, I’d still like to see ESRI leave the development of clients to other organizations and focus on making AGS as fast and scalable as possible.
89 Jamie // Mar 1, 2008 at 10:19 am
@Tim: I couldn’t agree more that ESRI’s should put their focus on the core API, be it the Web.ADF or ArcObjects. I would argue that they have done a very good job of this in the past and 9.3 promises more of the same. This does raise a larger observation I would make about ESRI: given their share of the GIS market, they are in the unenviable position of having to be all things to the entire GIS community. As I mentioned in a previous post, the underlying Web.ADF is fast and stable but, because of the overhead required by the map controls, implementation off-the-shelf can be tricky. There are certainly exciting new developments at 9.3 (we’re particularly interested in the availability of REST services) but this challenge for ESRI will remain.
The lack of off-the-shelf printing with the Web ADF is a good example of why ESRI is in a tough position. For a GIS product, exporting maps seems like a must-have for any web implementation. However, if you dig deeper, you realize that there isn’t any easy or efficient way for to build a print tool for the Web.ADF that can take into account all of the possible ways a map could be rendered in a browser. This leaves the developer community to build solutions that address the myriad of ways to tackle this problem. And this brings me back to your argument, which is that developers should be balancing functionality, performance, and usability for their specific end users and ESRI should continue to focus on building the underlying architecture that we all depend on.
90 Jamie // Mar 1, 2008 at 10:29 am
@Curious George: I’m also excited about previewing the new Web ADF developments at the dev summit and hope to see you there. But as our controls already support triggers in an update panel, have a robust client side JavaScript library for the current version of the ADF (to see our tool in action, check out the free version we submitted for the code challenge - http://arcscripts.esri.com/details.asp?dbid=15473) and are in production with our clients, we don’t have wait for the beta test this month or until a release later this year to worry about what might or might not transpire with the ADF. That’s not a knock on ESRI’s tools – we just take advantage of being able to choose our battles (per my previous post) and believe it benefits everyone, including ESRI.
Leave a Comment