[wdmmg-dev] OpenSpending Process (senger at rehfisch.de)
rufus.pollock at okfn.org
Thu Apr 21 16:42:57 BST 2011
On 21 April 2011 16:40, Friedrich Lindenberg
<friedrich.lindenberg at okfn.org> wrote:
> On Thu, Apr 21, 2011 at 5:18 PM, Rufus Pollock <rufus.pollock at okfn.org> wrote:
>>> On Thu, Apr 21, 2011 at 4:44 PM, Carsten Senger <senger at rehfisch.de> wrote:
>>>> We want to encourage contributors to write a custom loader where necessary.
>>>> It's the alternative to "Transform to OpenSpending Format2 for the "Custom
>>>> Loader" process.
>>> Hm, I disagree: I feel it would be much better for us to converge to
>>> using custom loaders as little as possible (and we're almost there
>>> already). Keeping the loaders up to date is increasingly a burden on
>>> our development and I'd like to get rid of the simpler ones in favor
>>> of a set of mapping descriptions. What do you think?
>> I'm a big +1 with Friedrich here. I actually think even our own
>> loaders should switch to producing our 'OpenSpending data' format and
>> we then load that in a standard way (the great benefit with this is we
>> now have a nice 'standalone' version of the data to redistribute to
>> others, to register on ckan.net etc).
> Paddling back a bit: I think there will always be data which we'll
> have to use some violence for in order to squeeze it in. As an
> example, IATI has schemes of classification for which any number of
> values can be specified per entry. Maybe we'll one day have this as a
> requirement but for the moment its clearly something we would hack in
> in a specific loader.
Not sure why this necessitates a specific loader (couldn't we go for
our 'json' in csv option ...)
More information about the openspending-dev