It's a goldmine, there are 350k datasets available, and a lot of them that will not make some people happy (criminality by geographic zone for example).

It opens possibilities to create a really nice multi-factor (e.g. crime, pollution, education) "best place to live" application.

A lot of the data is available in a very manageable CSV, sadly some is only available in XLS that may be a bit harder to manipulate programmaticaly.

For the ones that aren't formatted randomly in Excel:

library(RODBC)

channel <- odbcConnectExcel("myExcelFile.xls")

myDataFrame <- sqlFetch(channel, "Sheet1")

odbcCloseAll()

write.csv(

   file = paste("myExcelFile", "Sheet1.csv"),

   myDataFrame,

   row.names = F

)

For XLS, be sure to check out Roo http://roo.rubyforge.org/ or use Java libraries and JRuby.

for a pretty good python library:

pip install xlrd

This is a lot more data than I expected for a launch! I guess they wanted to get a lot of stuff organized before going public with it. Looks like some useful data that wasn't previously publicly available (or available under a free-enough license), from some brief skimming. There's some discussion on the French OpenStreetMap list about using it in semi-automated fashion to improve French coverage on OSM, especially for things like landmarks and landforms outside of the major cities, or adding metadata to existing landmark nodes.

Good news, for once french gov shows the way.

Or maybe just follows the pack. US, UK, Canada and others have been doing this for a while.

Italy is doing the same but as usual, too late and too slow: http://www.dati.gov.it/

It is a step in the good direction, but there are still too many datasets where the best available version under an open licence is from the US (that's often the case for french geographical data).

Well, the US is actually the exception regarding GIS data; not France.

That's true, and for more than GIS data, but it's nice to get alternate sources. Should improve both coverage and reliability, as well as reduce some systemic biases caused by which data is most available. For example, it's awkward that so many Wikipedia articles can only be illustrated with a photo of a non-American politician shaking an American president's hand, simply because no other freely licensed photos are available.

it's a step in the right direction but why isn't there an API ? seriously