[wdmmg-dev] [ckan-dev] Started work on the 'webstore' (a datastore with web API)

[Nick Stenning]

On Tue, Jul 12, 2011 at 12:14, Francis Irving <francis at scraperwiki.com> wrote:
>
> That's a bit timid!
>
> Surely the whole benefit of using an SQL database is to allow arbitary
> SQL calls in URLs, and the ability to call them directly from
> Javascript?

I'd just like to submit a strong +1 on this. SQL is a well-understood
standard for a good reason, and in my opinion the only arguments
against doing this are invalid here (security isn't an issue if we set
SQLITE_OPEN_READONLY for non-owners, and it we can kill queries that
take too long to run).

[Aside: you could *even* consider providing no SQL write support at
all, could you not?]

I'm all in favour of abstracting the simplest single-table features of
SQL onto a REST API, but where SQL really comes into its own is in
joining complex relational data. I'd argue that if you allow raw SQL
queries, there's no need to provide any JOIN facility from the REST
API whatsoever, which should simplify your lives a little.

I would *really* like to be able to store cleaned data in webstore,
and specify a (webstore identifier, SQL query) tuple as an
OpenSpending source.

Regards,
Nick