Quantum GIS

GS: Tyler, welcome to the next in the 'qgissers' interview series where we interview QGIS developers and users to find out their thoughts on QGIS, life, the universe and everything else we can think of to ask. Can you start by giving us a brief introduction of yourself?

Well, I'm relatively young, have a family with three kids and a wonderful wife. She puts up with my open source GIS addictions, which shows she is extra special! I live in Western Canada, British Columbia to be more precise, Prince George to be exact. I work for a great company, Timberline Forest Inventory Consultants.

I got into the whole open source GIS gambit about 5 or 6 years ago. When I started a GIS Specialist job, I got frustrated because I couldn't take the mapping and GIS tools home with me. The path from there led deep into the heart of products like MapServer, GDAL/OGR, OpenEV and PostGIS. I'm now a self-proclaimed user advocate for the community. My eyes are slowly turning to the desktop GIS side of things. Enter QGIS.

GS: Looks like you've been around the OSS GIS world for a while. Can you give us your perspective on its roots and where its headed?

As I dug into the roots of OSS GIS, I found a tremendous group of programmers. I was really encouraged to see so many Canadians involved with the tools I was using, that was a bonus. When you look at the core of many of the tools, you see some of the same data access libraries over and over again. Without these libraries, we'd all be re-inventing the wheel. Special credit needs to be given to the libraries Frank Warmerdam has produced, GDAL/OGR. By providing a common framework for raster and vector data access, products have been able to focus more on functionality than on data access. That's a big win for us. On the flip-side, however, it means that many more projects can start independently of each other. Whereas without GDAL/OGR we might have had to work together a bit more. We have a rich set of tools, many developed in isolation of each other. Going forward from here, many expect (and hope) to see some consolidation of tools. This is particular pressing on the desktop GIS side where there are a handful of projects, many not being actively developed. The spatial database backend is a critical component going forward. Tools that leverage the power of PostGIS, not only as a data store, but as a GIS engine, will shine above the rest.

GS: Do you think there are too many players/packages out there for a prospective user to choose from?

Yes, but not too many like with Linux distributions. That said, I'm all for freedom of choice and certainly appreciate any and all diversity. I deal with real, new users entering into this environment and many are perplexed. I believe that many come into it just looking for a tool to do one task, then they realise that with a bit more work they can have all the tools they need. It's addictive, but often draws new users, say GIS Analysts or hobbyists, into a programming environment. Heck, I never thought I'd be compiling from source! For the past couple of years I've been just trying to help bridge those two......worlds. More time is needed to help solidify products, many of which might not even be at version 1.0 yet.

GS: What do you think is missing in Open Source GIS?

This is a tempting question to answer - I will hold off on my global domination plan, and stick to some of the public items that I can share with you.

Bridge-ware.

I see two fronts that we are attacking: one on the product development side and the other on the public perception/buy-in side. I know that in a year or two we'll have all the products we need, ready to go. The core products are becoming stable. Someone interested in running a completely open source workstation can be pretty satisfied, or will be soon.

The other front we need to address is that of public perception or buy-in. Targets are not to find replacements for existing software, but to _bridge_ the gap between pervasive, often proprietary, tools and the open source GIS world. For example, some great guys in France started a project to allow ESRI ArcMap to connect to PostGIS. This project become PgArc. They released it knowing that if there was a good bridging tool available then 'traditional' users who are on proprietary platforms could start using the power of open source spatial databases in their day-to-day work.

Another example of a bridging tool could be the GDAL/OGR utilities that allow us to take proprietary data formats and convert/access them for use in our open source products. These tools help fill in the gaps between 'typical' users tied to a particular platform and the new or upcoming technology that will change their lives forever!

GS: So by Bridge-ware you mean the general concept of interoperability of tools rather than some secret project you have going?

Yes and no. A better medicine needs some temporary marketing. Perceived side effects of new drugs will scare some people away. Marketing is important to nullify these myths or at least clarify the truth. So it is with the OSS GIS world, but here the marketing is done through _real_ tools and applications being made available to new users. This is so they can try out the stuff before swallowing the whole bottle. The down-side to this is that, ideally, some of the bridging tools will become obsolete. Once everyone is over the bridge, you don't need it anymore.

I can't tell you about the secret projects. I mean I know nothing.

GS: Do you think the PgArc project was one of the reasons for ESRI's release of an interoperability product?

No, I don't think so. I guess I have to address my views on interoperability in their products first. I find it disappointing that interoperability isn't part of their core products. On one hand I know they need to focus on the real features of their products which means they can't do everything well, but on the other hand they've talked and talked about interoperability for so long that it is slightly disturbing to see them have to buy a solution from someone else.

I've been familiar with Safe Software's (safe.com) FME product for a long time. It's great stuff. I was excited to see them add in PostGIS support (for data conversion) early on. It was great. Then they created their ArcGIS connectors which allowed ArcGIS to access FME data sources. They didn't have a huge focus on OGC standards, but it allowed some neat data access that ArcMap didn't allow natively.

Anyway, to make a long story short, ESRI and Safe developed the interoperability extension collaboratively. We know, though, that Safe had the guts of it all already.

From what I understand, it doesn't support all the OGC standards (but I wouldn't be surprised if it does, the guys at Safe are great). Why there isn't robust OGC standards support in ArcGIS, I don't know, but it is frustrating. While the interoperability extension helps interoperate with other formats, I had hoped to see OGC support built into ArcGIS core, especially with all their touting......of the 'open' flag and their involvement in OGC. Sorry..you touched a sore spot.

TS We want to make QGIS a product that harnasses the best that OSS spatial applications have to offer to make the mythical 'everything but the kitchen sink' GIS system. Based on your observations of the OSS GIS world, what tools and libraries are out there that could be incorporated into an application like QGIS to make it a 'complete' GIS?

I know QGIS already touches on some of these, but I will mention all my ideas anyway.

1. Format support: GDAL/OGR and PostGIS read and write support are key. Then you can focus on other great stuff and let Frank [Warmerdam] take care of the data access needs.
2. Basic cartography: map layout capabilities and, as a bonus, some sort of mass production method. Even the commercial guys don't do this well, but this is the downfall of desktop GUI-based GIS products. For example, ArcInfo Workstation is great for automation, ArcMap certainly isn't. Perhaps what you need is a good scripting interface, or some graphical tools. I'm not sure.
3. Spatial analysis/manipulation: GIS functions for overlaying/aggregating features, buffering, etc. The key here, in my opinion, is that you do not reinvent the wheel, but harness PostGIS capabilities instead. Many use PostGIS as a simple data store, but the power is really in the geometric operations it supports. Aim for an interface to PostGIS geometric functions.
4. Complex table support: again, one area where commercial vendors crash and burn. When there is anything other than a one-to-one relationship, GIS fails miserably. Dumbing down relational databases into a 'flat' single table always makes me squirm. There has to be a better way. If QGIS can make use of tabular relationships, i.e. defined key relationships, in databases and give access to the user for complex querying, that'd be great!
5. The shapefile to PostGIS import tool should really leverage all OGR input formats. That would help me out a lot in my day-to-day work.

GS: We see most of the major software companies "dabbling" with OSS. Do you think this is an honest endeavor or the result of other motivations?

I read a good book called "Rebel Code", the subtitle is something like "A history of open source software and the linux revolution". It chronicled some of the early adopters of the OSS business model. Many got into it out of desperation - slipping market share, for example. Only a couple go 'into' OSS because a better product existed in OSS. It's always hard to drop your own product and jump on the bandwagon of another, particularly if you are selling your product. I'm actually a bit out of touch with what commercial vendors are doing with OSS these days. I can guess, though, they see increasingly mature technology and see it as a potential threat so they get involved somehow. I remember installing Oracle and seeing that they shipped Apache. That made sense. Of course Oracle wasn't going to develop their own web server. Are they into open source, definitely not, but they benefit from it and so do their customers. I see more of that happening, but I don't see a lot of other major vendors contributing back to open source. I guess, back to your question about motives, using software for the benefit it provides is honest. Posturing to make it look like they appreciate the product is dishonest. The companies that try to meet all markets by having two business models, will fail. You're open or you're not. I'm not a fan of the model that sells software as a product, because competitors can always mimic and out do your product. The services business model certainly appeals to me though.

GS: Lets shift gears a bit here and talk about your GIS habits for a moment. What GIS software do you work with on a regular basis?

MapServer, GDAL/OGR utilities and PostGIS are the big three in my world. You'll notice that there is no desktop GIS in my list, I'm tending to use web pages as my desktop lately.

GS: What do you think of QGIS? Be brutal - be honest - we can take it

I'm not alone in thinking that QGIS has the potential to be the best OSS GIS desktop tool. Let's talk about the pros first. I came to it because of PostGIS support. From what I gathered, it was one of the first products that even knew what PostGIS was. Then I was surprised to see that the product was actually nice to look at!

The WIN32 port renewed my interest in QGIS. I had given up interest because I couldn't run it on all the platforms I wanted to (Windows and Linux). But then along came WIN32 port out of right field! What the heck - and it worked.

I came back to QGIS and then found you guys had been going plugin crazy! These included some nice cartographic plugins too. I was impressed.

I'm not a C programmer, so I know I'll probably never be able to contribute to the project. I hacked around in the plugin template code, but that was during the days when the the plugin environment was being reorganized, so I gave up on it. Development has been very slow. This isn't so much a problem, as it is an observation.

I'm all for slower development that is this stable and this feature-rich. I'm having a hard time thinking of negative points though. My feeling in general is that QGIS is well positioned to lead in the desktop GIS world, particularly with the PostGIS editing capabilities. I've believed for a long time that the first product to support full read/write for PostGIS will grab the attention of many people.

For example, many of us store our data in PostGIS for use with MapServer. We then need a tool to help do edits and even for loading PostGIS. QGIS can help fill this need.

GS: Unfortunately the developers are an all volunteer force so our pace of development isnt always as we would like it. Do you think its important for QGIS to support Linux, Windows, and OSX?

Yes, I do think it is important. The most obvious reasons are portability. Some people want to run the same app on all their machines. One perspective that we don't talk about much is how the programmers look to the public. For example, when I see a project that has programmers who go through the process of maintaining a Windows, Linux and OSX build environment, I think they must be concientious programmers. Or maybe they just have too much time on their hands! Seriously though, I believe that broad platform support means an increased likelihood of more users in total. More users means more long term stability and confidence for users, knowing the product will be supported.

TS: You mentioned that you use MapServer / web interface for most of your GIS needs lately. What is missing from QGIS to make it your day to day GIS?

Simply put, it is need. I have little, but increasing, need for desktop tools. I use OSS GIS at home, but less so at the office. This is changing now that I'm in a management position at a different company. Traditionally, I haven't been doing production mapping work which starts to need some more desktop-like tools. Instead I've been supporting a handful of non-GIS corporate users with commercial products I had available to me. This is changing though, as I aim to decrease licensing costs and centralise data management in the company I'm at now.

More and more non-GIS users are coming on-board needing analytical and mapping tools. QGIS will be the first tool I push in. Is full read/write PostGIS supported at present in QGIS?

TS: Read/write support is being worked on and new editing features will be in version 0.7

When that is available, my first set of requirements will be complete. I will be the biggest promoter

TS: I noticed that you run an O'Reilly blog...what kind of involvement do you have with O'Reilly?

I'm just wrapping up a book that will be published by O'Reilly. The title will be something like "Web Mapping Illustrated - Using Open Source Tools". There is a focus on MapServer, but also a lot on GDAL/OGR utilities, OpenEV and PostGIS. It's almost a history of the knowledge building I've went through personally over the past few years. It was a direct response to increased 'newbie' demand for help. I was answering the same questions over and over again. My hope is that this book will get people up and running with these tools much quicker.

I write the weblog when I have time and to help demystify some of the technology.

TS: Do you think OSS GIS lacks an enigmatic figure head like Linus Torvalds? How can we get OpenSource GIS out from under the radar?

I have been promoting these tools as much as possible but find I hit a bit of a stone wall. I don't believe that the 'figure-head' is really needed, at least not at this point. The GIS group is a different kind of animal, you just need to feed it tools and it will follow you anywhere.

Most (normal) GIS and mapping types don't really care about how a product does what it does, but many of us obsess about it. The concept of "free" tools for GIS gets some attention, particularly when you compare with current commercial price-tags. Beyond that, the idea of working together to develop tools or to improve a product, are quite foreign thoughts to most of the domain. For many it is more important to be using the 'latest' technology espoused by the major vendors. There is a comfort with the "same old" and an apathy to something better.

I believe that the way to get out from under the radar is to focus on markets and not on the concept of OSS GIS. For example, I was just at the Geotec conference in Vancouver last week. We had a special session on Open Source GIS. There were five of us speakers talking about success stories, the business cases, etc. We had 65+ attendees, that was encouraging. We don't get to talk to a lot of the people because they are interested in their domains, their marketplace, centered on their expertise.

Unless we go to the "GIS for Natural Resource Management" sessions and present things, some people will never see them. Likewise, we need success stories to be shared at other, non-GIS venues. In essence, talk about problems and solutions rather than having the technology or licensing as the focal point.

TS: QGIS is in a rapid (some may say rabid :-) growth phase. Each version brings us that much closer to a functional desktop GIS that is easy and approachable for non-technical users to work with spatial data. More users equals more demands, more bugs to fix, more questions being asked on mailing lists and more time being spent by developers on administrative tasks rather than coding. Other projects have obtained 'corporate pickup' because they have demonstrated themselves to provide viable solutions to big business. We want to reach that point where businesses think 'hey QGIS, that will save my bottom line on desktop GIS licensing'. Who is our potential 'IBM' or 'Oracle'? By this I mean, do you think there is gonna be a GIS version of the OSDL (Open Source Developement Labs) where sandal wearing bearded (Q)GIS developers will just get paid to hack by businesses because they know we will generate 'cool stuff' thats going to be useful to them further down the line?

Don't bet on the behemoth adoption program. Pick a domain and aim to saturate it - or at least annoy it. When you've captured the needs of that domain, then move to another. The vanilla approach to GIS is necessary at first but real users look at real functionality for their (very real) probleme du jour. I don't believe there is any rocket science behind QGIS, but the other big guys 'looked like' they had rocket science when they were introduced.

Those days are passed. Innovation will repackage old technology so that it better applies to a particular set of problems by a group of people.

Let's use word processing as an analogy. OpenOffice matches features of other commercial products pretty substantially. So why don't people switch? So would make you switch? Would a very convincing piece of marketing do the trick? I doubt it. However, when you have me telling you all the time that I use it, then you might start to consider it.

The problem with the GIS market is that it is already held in a monopoly. Unless the monopolist is ready to adopt QGIS, who are the great big guys? They're too invested in their own technology and can't abandon it. Even if just for the sake of their users. But I know a handful of people in, for example, forest management who have need for basic desktop mapping capabilities. They don't have the budget for purchasing anything, because the cost/benefit isn't clear enough. Give them QGIS and they will be ecstatic to have the tools they need. They don't care who made it. They will tell all their colleagues about it though, and if I'm savvy I'll have it packaged on a CD ready to give them when they stop by to learn more!

In short, it's a user group focus and not a corporate adoption focus that will carry things forward. All that to say that a user group could be a large organisation with buying power, like a government agency.

TS: What is your problem domain? Natural resource management? What is it you actually 'do' with GIS?

I spent the most of the last decade using GIS as a data management tool in the forest management domain. I worked for a forest products company (Lignum Ltd.) who managed timberlands. Part of the company's mandate was to apply innovative solutions to sustainable forest management practices. This included funding research projects and hiring consultants to analyse scenarios using tree growth models, forest fire models or forest pest models, etc. All these required base mapping and GIS data that I helped shuffle around. I have recently joined a "forest inventory consulting" company (Timberline Forest Inventory Consultants) who can provide some of these innovative solutions to the industry. Many companies use GIS tools to do mapping, and never use any of the analytical capabilities that really embodies GIS.

We are no exception to this rule, although we do more analysis than most companies will ever do. This is because we get hired by companies, some of which have internal GIS departments, to do complex analytical tasks that are our specialty. Our company has experts in sustainable forest management, GIS, Ecology, Forest Inventories and resource analysis. My department, GIS Services, provides data management, analysis and mapping expertise to the rest of our company and to external clients. These clients are not all forestry, but will include government, regional government, academia...a range of clients.

TS: So what percentage of your application usage is OSS GIS based?

Before I started: 0%. I was hired, with the full knowledge of my advocacy of the OSS technologies. Part of my job has been educating the others in the company that OSS alternatives exist, not just in GIS, but throughout our broader corporate toolsets.

TS: So you are not employing the traditional approach of bringing OSS software in through the back door?

No, not at all through the back door. When in industry, there is little benefit to researching and developing new solutions because the benefit to the bottom-line is unclear. But in consulting there is always a benefit. Costs come up first, but pale in comparison to harnessing innovative technologies to meet our client needs quicker. The bonus is that we can then give them more tools to make full use of the information we provide. For example, traditionally we would run some analysis, print a map and be done. Now we can run the analysis, print a map and post the results on a web-map for them to use interactively.

TS: What are the three 'must visit' web sites you visit every day?

Man, I don't have a lot of regular web-page habits.

• http://maptools.org - a growing portal for OSS GIS products
• I'm into the PostGIS manual a lot: http://postgis.refractions.net
• And maybe http://bugs.microsoft.com ? ;)

I actually use my own Global web map a lot: http://spatialguru.com/maps/apps/global

TS: What emerging OSS GIS technologies should we keep an eye out for?

If anyone asks this question, I usually answer: QGIS. It is really the closest option to filling the desktop GIS need that I see at present. One other thing to watch out for is the development of some sort of "Foundation" to support OSGIS development and community. This could be similar to the Apache Foundation. This will be critical for moving forward. It will provide clients/corporate users some level of assurance of longevity long after the current developers are gone. It can also help bring in funding, but not so much as a mandate. Rather as a professional front to the disparate team of developers. It will be a support network and professional front to particular products.

For example, if QGIS is adopted by a corporation. How will they fund development? Write a cheque directly to one of the developers? That doesn't help them feel like there is a large group of developers working on this. A foundation can help funnel resources. It can also help maintain essential resources, like web site hosting. Right now, for example, the University of Minnesota hosts the MapServer web site. It works out well, because they are committed to it. In a worse-case scenario, the administration of the university changes and dumps this project. What next? A foundation would be a non-profit organisation who is solely responsibly for the life and image of the project.

If I could add one more thing to watch for. I'd like to note that a critical conference is coming up June 2005. It is a carry-on of the MapServer Users Meeting and Open Source GIS conference held in the past two years. This time, it will bring together some of the other key open source GIS communities from around the world. I believe there will be folks from the GRASS GIS project and from EOGEO and other groups there. This will be the first time some of these people have ever met and it should be a great synergy to the whole effort.

http://mapserver.gis.umn.edu/mum/mtg2005.html

TS:We would like to thank you very much for taking the time to answer all our questions!