Sunday, February 12, 2006

Slow progress

Well, it's been a while, but I've been making slow progress. I decided to go with QGis and GRASS because I can't find what I want in other packages. That's not to say that they're not there, but the features I'm looking for are not obvious to me, except in QGis and GRASS where I'm pretty sure I'll find what I need.

Something else I've discovered in the course of a few work-related meetings in the past week; ESRI ArcGIS is giving me a vocabulary in GIS that others seem to have trouble with. I'm also finding that the tools I'm finding easily in ArcGIS are not so easy to find in any other solutions because the ESRI names are unique. Maybe just working with the other software will expand my vocabulary.

Anyway, I'm finding some awkward bits in compiling everything needed for QGis. The required software includes GRASS if you want to use it, and GDAL. Problem I'm struggling with is that if you want the two to be aware of each other, they need to be told that at compile time. Which do you compile first when the other doesn't exist? I'm compiling both without awareness of the other, and then re-compiling. I'm struggling to get the compilations to complete, so I haven't been able to get to the point of testing to see if that approach works. Haven't found any documentation that helps.

One issue in compiling is that my latest FC4 update installed a newer version of fft utilities that GRASS doesn't appear to help. Once again the open source model kicks in; I'm subscribing to every listserv for each of the packages in the hopes that someone is willing to answer newbie questions that may or may not be worth answering...

Ever catch yourself checking the e-mail every few minutes on a Sunday in the hopes that a group of developers is actually hard at work?


Ben Slater said...

Good luck! I'm trying the same thing on my home PC, though with PCLinuxOS as the operating system, and without GRASS. I wasn't sure what to do about GDAL and GRASS either, so I just dumped GRASS altogether. I'm still working on compiling QGIS from source - I get a segmentation error whenever I do make. I'll be interested in seeing if it works out for you.

deer_smacker said...

Hi, it sounds like we are in the same boat. I am working on a project to redevelop a web-mapping application that was built with ArcIMS and SDE with OS software. I started using Ubuntu as the base Linux install but I noticed that there is a lot of information on the Mapserver web site about Fedora which has made me re-think my decision. I was wondering why you decided to go with fedora?

Alpargatus said...


Do you know gvSIG?

gvSIG is a tool oriented to manage geographic information. It is characterized by a user-friendly interface, with a quick access to the most usual raster and vector formats. In the same view it includes local as well as remote data through a WMS or WCS source.

License: GPL

URL: (in english too, click England flag)

Cresques said...

Hi Michael,
I'm working on gvSIG projecto, and I think it may help you. It's a under development desktop GIS (like QGIS and GRASS), and besides working with vector sources (SHP, DGN, DXF), it may use free source spatial databases (postgis, mysql), as well as working smoothly both with raster files through GDAL embeedded library, and with web services datasources (WMS and WCS). In a couple of days will be available our next release preview, adding support for OGC and z39.50 catalogues and a brand new and powerfull WMS client.
It works quite well both on windows and linux.

matt said...

Hi Michael, I too am very familiar with ESRI tools, specifically ArcInfo Workstation and ArcMap and have spent some time investigating open source alternatives. There are a lot of projects out there that call themselves GIS but upon investigation turn out to be merely spatial data explorers (though Google Earth has proved that "merely" an explorer can be pretty darn good.)

You've already discovered qgis and grass, so I'll pass over them. There are only two other libre software packages I consider worthy of time if you actually want to *do* something with data: JUMP and SAGA. (

SAGA is new to me. It's feature set appears to be fairly mature, oriented more to raster than vector processing. At first blush the UI is old school with lots of little panels and such. However it is clear that thought has been put into using the interface as there are lots of small nice touches. For example changing the sort order an attribute table dialog mirrors it in the selected Identify panel. Also the default Zoom mode is left click to zoom in and right click to zoom out (no middle-click to pan though). In a quick 15 minute trial I couldn't figure out how to add/delete shapefile attributes or values.

Saga is worth looking into in more depth.

JUMP - So far this is the only one which offers edit and analysis capabilities out of the box, and except for grass it is the only one I've seen actually mention topology. With the OpenJUMP variant I was able to load and edit shapefiles both by attributes, including changing the schema, and by geometry in less than 5 minutes.

I couldn't figure out how to merge multiple shapefiles together while retaining their attributes (maybe 'cause the schemas didn't match?). However if all I needed was the geometry it's a simple act of selecting multiple layers in the table of contents and then copy>paste into a new layer.

Excepting GRASS and SAGA, Jump is the only one which I would actually consider a GIS (qgis is bumped off the list because it can't do any editing -- yet).

Tim Beermann said...

I have been doing similar work with a different approach. I am using Google Earth as the end user interface, and MapWindow GIS as the data converter.

Since my primary data format are shapefiles, I have been using other GIS packages like JUMP and QGIS for their respective editing strengths. MapWindow GIS is then used to cartographically render the shapefille to KML for viewing in Google Earth.

I would eventually like to make Google Earth the primary interface, using PostGIS and a couple of different open source GIS components for the data and business logic.

Michael said...

I work with Fedora because I started using Linux in 1998. At the time I was working for a university, and was also supporting HPUX machines. RedHat Linux was the flavour used by the research group as a possible replacement for HPUX. It was also the system used by a couple of mentors who guided me through the process of building a 48-node beowulf cluster.

I had a ball with that project, and have not had any strong reasons to shift away from the RedHat/Fedora releases.