There’s something to be said for blogging when something’s fresh in your mind. (Sorry Glenn, not even close to real-time on this one) And no, it’s not because I had too many Crown & Cokes. After a few days away, my family needed more attention than my mistress (though her big CRT was glowing seductively) and time plays some interesting tricks with memory. I’ll do my best to give you the highlights though. If you haven’t read about Day One or checked out the conference home page yet, do that first. One thing that I do remember is giving Matt a workspace that will likely give him nightmares, and make him wander around muttering about crazy users.
The morning featured Don and Dale giving a “Road Ahead” presentation in their casual funny style, complete with some really bad puns. One of the most important items that I got out of this were their plans to deal with the next hardware tidal wave with support for 64 bit operating systems and multi-process capabilities to take advantage of the soon to be ubiquitous multi-core CPUs. I wasn’t entirely clear on how the multi-process support was going to be implemented, but it sounded similar to the way that Apache sets up multiple children to listen and respond to requests. As well, there were plans for additional formats as well as improved support for existing formats. One of the most exciting items here was GeoRSS support. I know that this might not sound like a lot, but imagine using GeoRSS as a web-based data synchronisation tool (insert/update/delete). Yes, I suppose WFS and Filter could be used in the same way, but GeoRSS is new and trendy. :)
Rich Geometry is another important area, as it allows for the preservation of nasty items like a linestring with an embedded arc. This is a godsend for anyone that has to deal with CAD data on a regular basis. I spoke about Python improvements in the previous article, but I just noticed that there’s a recent story related to this on the Safe blog: Vancouver Python Workshop and FME. Raster support has seen an order-of-magnitude improvement in performance as well; a projection task was shown that was reduced from over an hour to less than two minutes in processing time. This is great news. Now I can put in some unreasonable requests for improvements in the raster processing capabilities. First I’ll have to apologize though. I interrupted the presentation and made a rather nasty comment about one of the 3rd party integration products that, while there are strong reasons for implementing it, doesn’t hold a candle to FME Workbench for usability. I wasn’t nearly as tactful as I should have been, and I hope that I didn’t scare away any potential users. Anyway, I’m sure that I missed some of the new features; hopefully the slides from the conference will be out by the middle of next week some time.
After their presentation, Dale and Don had Dmitri come up on stage and do some demos. I’m a bit fuzzy on these, but I am pretty sure that they were on the new third-party transformers. These are sold by FME, but are developed by other organisations. MRFCleaner (developed by MRF Geosystems) allows for some very complex automated quality assurance and cleaning. Many of the functions (overshoots, undershoots, intersections, etc) supported by MRFCleaner could be managed in FME previously, but they take countless transformers and are a real nightmare to maintain. I am seriously considering adding MRFCleaner to my licensing.
ArcFitter (I didn’t catch the developer’s name) CurveFitter (developed by Tom Inloes of TCI Corporation) allows you to start recovering from the legacy of GIS data models that only supported points, lines, and polygons by retrofitting your stroked curves with fancy new arcs. As well as being a better representation of certain features, this can give you a reduction in storage space, especially in heavily splined curves.
After this came a set of presentation sessions. I was planning to go to the Best Practices for FME talk, but instead ended up sitting with Tom (one of the developers) chatting about KML 2.1 support. Warning: it’s not safe to sit down with me and talk for any length of time. I might just get around to discussing my thoughts on how open source geospatial will cause some proprietary vendors to reconsider their core competencies, and others to reorganise as their bread-and-butter components become commoditised. I believe that there will continue to be a place for well designed, efficient, and innovative products like FME, but that a lot of the fat in the industry will be trimmed.
Lunch was again good. I don’t know if I’ve said this, but the networking at this conference was incredible. I was lucky enough to speak with folks who were doing things as varied as: using FME to aid in moving away from a VAX/VMS-based system, building financial asset reports for taxation purposes directly from drawings, and using FME’s geometric processing capabilities to build complex cartographic output. There are not enough hours in the day to do all of the cool things that FME makes possible, so it’s nice to get some vicarious exposure to others who are working on them.
After lunch came the lightning talks. These little five-minute show and tell pieces were great fun, accompanied by computer-generated lighting sound effects, and a virtual audience that would clap you down if the thunder wasn’t loud enough. This was a great format to learn a lot in a little while, and I hope that this tradition continues in future conferences. I think that if they can get enough volunteers, they should allocate even more time to these tidbits.
The lightning talks were followed by another set of user sessions. This time, I went to the FME and Cartography session. This was really interesting to me on a personal note, as I am currently banging my head against trying to join some Autocad blocks up to some Oracle Points, and bringing them into MapInfo maintaining colour, size, rotation, and other goodies. Not fun at all, but these presenters seemed to have done some pure magic in this area. Hans (an FME MVP) gave a great presentation on how he used FME to improve the cartographic value of a street map, while SRG (another FME MVP) provided a very tasty recipe for SVG a la banana. I hadn’t looked into SVG much before this, but it sure looks powerful and flexible. During this session, a fully symbolised vector-based topographic map was shown running in Internet Explorer. Cool.
I won’t say too much about the ice cream (yum), the closing comments, or my request for flashing lights and running numbers in Workbench. Suffice to say, I had a great time at the FME UC, learned tonnes of stuff and got to speak with a lot of really smart people. I also felt like a bit of a celebrity, which was really odd. My wife doesn’t want me to go to another FME conference for a while. She says my ego’s already large enough… :)