Comments on FOSGIS

Below find a comment on the article by Gianni. I do not necessarily agree (see my notes). But I guess the interested reader may find his own conclusions.

Dear Mr. Steiniger,

I just read your article "An Overview on Current Free and Open Source Desktop GIS Devlopments" and found it extremely interesting. I fully agree on most of the content of the article and I really appreciate the introduction with the clear explanation of differences between Free, Open Source and licenses. I also liked that you identified the university & research perspective, which I agree is different from an end-user and long-term one.

I would like to add some comments on the tables and arguments you presented; I am working in ESRI Italy, so my comments could sound as a "proprietary" point of view, but I think they are not.

1. Table 4
I think that this table is slightly misleading. What I mean is that only few elements are presented in the Functionality column, while it should be given also a measure of the number of functionalities available, not only a Yes/No answer. For example, in the vector/raster data rows, the GML3, GML3-SF, WKB, WKT, Tiger and many other format are important, as well as the capability to read (or write) some vendor specific proprietary format like DWG (not only DXF), MapInfo, Microstation, Intergraph's Geomedia. The same remark applies to Databases: what about SQLServer ? I know it's not a Spatial DB, but it's widely used and the Express version is free (in the sense of free beer, this time!) and many software can use it as a Geodatabase. What about linking different databases in the same map ? And versioning and replicas mechanisms ? These, and many others, are very important functionalities not only form a user's point of view, but also at research and university level, because they use concepts that are fundamental basis not only of GIS but of many enterprises systems.

[sstein]: I just picked the most important exchange formats - i.e. those where the specifications are avilable. What is the sense of listing every xth format - shouldn't we rather strive to work only with a basic set of formats? On the databases.. same thing: I just picked the once that I see relevant, i.e. most used.

2. Table 5

• Why "support of open standards" is an advantage only for FOSS ? ESRI (and other commercial companies) people works in OGC and ISO and W3C teams and for this reason standards are often implemented in product even if they are not officially released.

[sstein]: ok.. this is a bit difficult to answer. But in general big companies have a tendency to make things as much standard as they need to - to get a shiny certificate. I think it is kind of unbelievable how the OGC standards are setup, so that almost every provider gets compliance. So.. my very personal oppinion is, that the OGC is not realy working the way it should be. On the contrary - open source is "open" - so the ease of use and adoption is given and a stanard can evolve.

• Why customisation at API level only on FOSS ? ESRI has a wide library of C++ classes (ArcObjects, also in Java) with which a user can build a new application from scratch, and in fact ArcView and all other products are built using ArcObjects.

[sstein]: right.. but here we are talking about a specific "product" and not openenss of source code. ESRI provides an API - but do others? And what is meant (sorry for being mistaken) is that you can build your own API functions.

• In the disadvantages I don't agree on the "support only as long as software company exist", since I had experience on some OS software that as been abandoned: the "critical mass" for an Open Source project to exist equals the widespread of a commercial product.
  

[sstein]: not sure, you mix up two things a) there will be no support if the company crashes [if the product is not overtaken by somebody else] b) Open source support will exists if people exists that are willing to pay for the support, because the "openess" enables anybody to provide support.

• In this table there are at least two important aspect that are not considered:
  
  • a. support for old versions - This is something that FOSS does not. Not even very large project like Apache community will give support on old versions of the software. The usual way is "upgrade to new version, where the bug is solved". Usually commercial software is committed to give support regardless on how old the version is.
    

[sstein]: see my comment above. If you pay it - you will get it

• • b. evolution- This is by far the most important issue. One very important aspect in GIS software is scalability, which means the possibility to move from low-level desktop up to Enterprise GIS, adding new pieces, without rebuilding everything at each step. This is something you can not achieve with FOSS: if you start using QGis or uDig, and you want to move to web publishing, you have to start using something like MapGuide, which is a completely different piece of software. This is expensive, and the final design is often not very efficient in terms of workflow and/or data duplication/transformation.
    

[sstein]: my two cents: What about the advantage of specialisation? Do I want to get all the things that I do not need? And, uDig and QGIS and others are customizable. So, just find one person that makes you a special edition with lots or less functionality for the type of workplace you need. Btw. QGIS got now a web-server component. Clearly there is evolution in FOSSGIS too.

Thanks and regards
GIANNI CAMPANILE
[sstein]: I thank you too.. as some of your comments made me think and are difficult to anwers [i.e. no black/white case]. However, I am glad that you agreed to publish your comments, so that people can get an own idea from our two perspectives. I.e. I may also be wrong with my comments on yours.

sstein: 17 Nov. 2008