[ckan-dev] Hello

Wed Dec 8 11:30:27 UTC 2010

(changing the cc to ckan-dev, best to have this type of thing in the
public archive)

* [2010-12-08 11:12:44 +0000] Seb Bacon <seb.bacon at gmail.com> écrit:

] Hopefully by now you know I've joined the team.

Hi Seb, welcome to the team! (I didn't know you had joined)

]  To get it to work, I've needed to upgrade SQLAlchemy to a
] recent version, despite the fact that it's pinned at a version <=
] 0.4.99.
] 
] However, that appears to break VDM.  I've just started looking into
] why, but thought I would drop the list a line to check if anyone has
] any information about the reasons for pinning SQLAlchemy to an earlier
] version. 

I had a look at this a while back when I was writing the SQLAlchemy
back-end for Virtuoso (http://packages.python.org/virtuoso). One of
our potential implementation strategies for LOD2 involves using
Virtuoso instead of PostgreSQL. Since SQLAlchemy changed the structure
of the back ends between 0.5 and 0.6 I was led down the path that you
are now treading.

At the time, I think I was able to get all tests for VDM passing with
versions 0.5 and 0.6. It may be that something (hopefully simple) has
changed between then and now. I imagine this will be simple to sort
out. 

More difficult, I found, was with CKAN itself. There are some places
which I think David Read may know more about that cause a warning
relating to use of polymorphism -- you should see these in the logs
running CKAN normally. Upgrading to SQLAlchemy 0.5 causes these
warnings to turn into exceptions. That's where I left it.

I think it would be extremely useful if you were able to get CKAN
properly running with SQLAlchemy 0.6 -- it would give us quite a lot
of flexibility with the database back end.

Cheers,
-w
-- 
William Waites
http://eris.okfn.org/ww/foaf#i
9C7E F636 52F6 1004 E40A  E565 98E3 BBF3 8320 7664