15

If you judge by the amount of the questions regarding FOSS software that are being asked in GIS SE many users seems to prefer FOSS over proprietary software.

I've read some articles - more precisely some personal blogs - supporting this choice. Furthermore corporate giants like Esri seem to acknowledge the open source development movement.

So, I am asking your position in this matter.

Why do you use open source tools?

What are the advantages or disadvantages of your choice, if any?

Improve this question
12

Reasons to use FOSS:

1) It's free!

2) Ease of access and capabilities - most basic remote sensing tools (e.g. filtering) are available with FOSS, so there is no need to pay for it

3) It's open - the algorithms you are using aren't 'black-boxed'

4) The ability to add/modify your own tools

also;

5) Telling people you use FOSS makes you feel cool

6) You enjoy encountering an assortment of interesting bugs and crashes while testing new beta releases

7) You like the elitism of being one of five others using the softwares' user forum

8) Unexpected results are novel

Improve this answer
4
  • 4
    I used FOSS software in my academic project two years ago for the reasons outlined above. Above all I did not want to use pirated software! (Universities don't issue free student licenses here). I always feel proud to say my research was accomplished with 100% FOSS4G applications. – Chethan S. May 21 '13 at 17:00
  • 5
    and in majority, they are multi-platform (not restricted to Windows like ArcGIS) – gene May 21 '13 at 18:03
  • 4
    In our view, the explicit use of Free and Open Source Software (FOSS) with availability of the code is essential for completely open science. For details, see Trends in Ecology and Evolution. doi:10.1016/j.tree.2012.03.009, Full text: tinyurl.com/tree-four-freedoms – markusN May 21 '13 at 20:17
  • 1
    "Professional" commercial GIS programs usually cost thousand of € and keep costing yearly maintanance fees. QGIS + GRASS + PostGIS + R + ... for free and with great community is much better investment to hobbyist or small company – simplexio May 22 '13 at 9:07
5

The bottom line for me is that I use the best available software for my RS work, regardless of whether or not it is FOSS. With that said, the FOSS tools I consistently use are R, Python and SPRING. I use SPRING for image segmentation out of necessity when eCognition is unavailable to me. In my opinion, R simply has no substitute for many applications related to RS (e.g. decision tree classification, spatial statistics, etc). Python is my link between the RS and the GIS worlds and used mainly for automation.

I find the commercial programming languages directed at RS, such as IDL and MATLAB, to be clunky and less user-friendly than R and Python. In a way, I feel as if many commercial software packages and languages are imposed on us rather than growing naturally to our needs via the open-source process.

Improve this answer
4

If you want to use a GIS and you are not in an academic institution or in a company witch have special conditions or money to pay for licenses, what solutions ?

  1. pirated software ?
  2. FOSS ?

For me, the second solution:

  • I can participate or have the illusion of participate in the development of the software in various ways
  • when I suggest something or propose a solution, if it is interesting for the community, it is directly integrated in the next version of the software, and if not, too bad, maybe another time.
  • I do not need to use Windows
  • I am not subject to the obligation to use "black boxed" functions
  • I don't need to upgrade by obligation
  • and most important, I prefer cooperation to competition

For me, that is the Free (and not the free of charge), and I am ready to pay if it was necessary.(and I use exclusively Python, QGIS, GRASS GIS, gvSIG, OpenJump, R and Octave (free clone of Matlab), PostgreSQL/PostGIS, SQLite/Spatialite, LibreOffice, git, gitHub, etc., all Open Source, not in Windows)

Consider the gvSIG development process:

  • gvSIG was started in 2003 when the Conselleria d'Infraestructures i Transports (CIT) of Valencian Country (Spain) proposed the development of a software for the management of the spatial data (as an alternative to ESRI products)
  • A private enterprise (IVER Tecnologías) develop the software together with the Generalitat Valenciana and the Jaume I University of Castellón.
  • It does exactly the functions required by the Community (no more, no less at the beginning)
  • but given its status (Open Source), the project has been joined by other regional and national administrations, various national and international organisations, programmers, that have added what they want, moving the project forward, towards a complete GIS (with 3D now)
  • and the software is available for Windows, Linux, Mac OS X and Android (gvSIG Mini)

Is this possible with a proprietary software ?

Improve this answer
1

If this is about 'industry' then you use whatever maximises profits (MS Paint, Excel, whatever). If this is about research in science, in extending the limits of human knowledge, then you have to use open source software.

Because otherwise you are doing alchemy.

Here is my data, a grey lump of lead. Here are my results, pure gold. I'm not telling you how I got from lead to gold, its in this bag. No you can't look inside.

Science requires openness - its not an option.

Improve this answer
2
  • 7
    I get it and am sympathetic. But clearly there's more going on here; the situation is complex. Otherwise how do you account for the very many peer-reviewed scientific papers published which rely on ArcInfo/ArcView/ArcGIS analyses? If you try to say the reviewers and publishers were mistaken, that would make you a crackpot and you will no longer be heard in the debate. Consider this analogy: what do you do when your science research relies on a measuring instrument manufactured by a commercial vendor who will not disclose some details of its workings (trade secret)? You calibrate it, of course. – whuber May 21 '13 at 18:32
  • 7
    I might argue that science needs open-source hardware too... – Spacedman May 21 '13 at 21:33
1

Being a researcher one should have complete access to source code how an algorithm functions. So proprietary software is not a good choice.

Improve this answer
1

Proprietary tools are like cars with a "welded bonnet" you cannot see whats happening under the hood, on the contrary open source allows you to view, study, modify and make locale specific adaptation (regional language support), improve the overall efficiency of the underlying algorithms etc.

Improve this answer
0

We consult in the tools that our clients want us to use. If clients want the maps done in Excel and MSPaint. Hell we'll do it there.

In all seriousness though, ESRI is an industry standard and it's difficult to be a consultant if you don't consult in its stack. On projects where ESRI is not a requirement, you can experiment with other tools.

On some types of projects, analysis and results oriented fields, I find that FOSS tools score better than proprietary stacks due to the community size and ease of upgrading or updating an existing tool

With more and more web implementation of GIS tools and technologies, I assume there will be a big shift from "Standard" to "Standards" meaning that we'll 4 or so leading products split in half between proprietary and FOSS

Improve this answer
0

Well I would suggest using open source in RS because you learn stuff!!! Otherwise complex algorithms are ensambled into wizards and the end-user does not know the complexity and richness of what is below that. Applying a low or high pass filter just because it is in an options menu does not elighten the end user.

If you use FOSS for RS you will learn things the hard way once and for all and not just a kitchen recipe for classification, NDVI, NBR or composites etc.

This applies to research and commercial from my point of view.

I took a RS course using ERDAS and it crashed quite a lot to a point where I really thought it was some sort of beta and they didn't tell us about it; the ortho-engine module just kept crashing with a spot 5 image.

In an unsupervised landsat 5 classification which I re-did later with GRASS, things were smoother with ERDAS but the result was pretty similar.

Improve this answer

Not the answer you're looking for? Browse other questions tagged or ask your own question.