[Grok-dev] Re: New Grok user with some questions
faassen at startifact.com
Wed Jun 25 08:57:11 EDT 2008
Kevin Teague wrote:
>> - I found some references about local utilities but no real
>> documentation about it. Is there some documentation or sample code on
>> how to use local utilities in Grok?
> There is a small mention in the Official Grok Reference, but the
> reference isn't finsihed and published on the web site yet:
> In the absence of documentation, the ftests and tests directories in the
> grok source package make for reasonable examples of how to do things.
> But asking questions on the list is also good, since it has a chance of
> prodding someone into documentating something :)
Yes, for now I think the ftests in Grok are most instructive.
In brief, there is a 'grok.local_utility' directive which you can use on
a grok.Application (or sites in general). When this site is created, it
will automatically create and install the local utility for you.
>> - One of my main problems with ZODB is how to implement m:n relations.
>> I think most people comming from relational databases have this
>> problem and a short how-to would be quite helpfull. Perhaps this would
>> be a nice lighting talk on the Europython 2008? ;-)
> lovely.relation (http://svn.zope.org/lovely.relation/) is AFAIK the best
> place to start for modeling many to many relationships (um, M:N to is
> many-to-many)? It's a Zope 3 package built on the lower level
> zc.relationship package. I've only gone far enough with this package to
> create a very simple relationship in Grok 0.12, but maybe once I've used
> it a bit more I'll write up a how-to - there is a fair bit of docs
> within the package itself as well.
I just mentioned my stealth package z3c.relationfield. I just realized
it is actually in Zope's SVN already. It only does 1 to 1 relations
right now, but it does index them and so on, making quite powerful
queries on the relations possible. It also hasn't been publically
released yet. 1 to many relations are still coming.
The nice thing about this package is that it is relatively simple to use
in an existing schema, while it still offers powerful indexing. That
said, I can't promote it too much yet, as it needs a bit more polishing.
More information about the Grok-dev