Grand Lyon Open data sometimes deliver updated versions of their
vintaged CityGML data. This means that e.g. 2009 data will be changed
which can break all our dependent treatment pipelines.
In order terms reproductibility requires tagged (well known and preserved)
data.
They are two ways to adapt to this situation:
long term run: discuss with the Grand Lyon Open data operators in
order for them to find a naming scheme (and the operational process)
for vintaged data suffixed with a release version/date.
short term run (because local):
download today's version of the vintaged data
place it, together with the corresponding md5 signature, on e.g. rict2
server this data with http
adapt our scripts to check the data we download from grand lyon data
against the signature. When some data was changed download our
identified copy instead and proceed with computations
manually assert that the new version of the data does produce the same
result and adapt accordingly (change the script, change the rict2 vaulted
data...)
Grand Lyon Open data sometimes deliver updated versions of their vintaged CityGML data. This means that e.g. 2009 data will be changed which can break all our dependent treatment pipelines. In order terms reproductibility requires tagged (well known and preserved) data.
They are two ways to adapt to this situation: