pcplantdb

[Richard Morris <webmaster@pfaf.org>]

Chad Knepp wrote:

> On a personal level things have really been going my way and I think
> taking a break from computers has been quite helpful... sorry to scare
> you Stephanie.  As I mentioned I am no longer a resident of DR and the
> tensions of leaving a community have diminished significantly since
> leaving.  Other significant events include getting a check and a very
> sweet encouraging letter from Scott (Thanks!).  As Stephanie said,
> Eden is no longer working on galatea.org because it moved to a new
> server and my ISP isn't really interested in supporting it.  There was
> also a couple of days where I wasn't receiving email because of
> assumptions I made and assumptions my ISP made about my email account.
> I've gone back and read list related stuff in the archives but if you
> sent me anything offlist that was important and it bounced back to
> you, you could/should resend it.  Short of a couple of days of
> gardening and bit of organizing/remodelling/domestics I have this next
> month entirely free up until Thanksgiving and will then be spending a
> week in Michigan with my folks.
>
> I feel very energized to resume work on permaculture.info and would
> like to talk about the direction that could take.  I am willing to
> continue work on HG/Zope if the group chooses this, but I still feel
> like the best choice is to start with Eden instead.  I've done a
> little bit of research, looked at a couple of ORMs and am currently
> not really in favor of them because with a relational backend you can
> either sacrifice some of the power of the relational model/technology
> or your ORM is not very object like.  I am currently in favor of the
> idea of focusing on the database itself.  By putting a lot of energy
> into getting a very well structured/correct (fully normalized)
> relational database we can have a dataset that is easily manipulated
> by a variety of software tools such as Eden, TouchGraph, custom
> clients, etc.  This is another point Richard was in favor of early on.

Yep I'm still very much in favor. To me it seems like the DB (or object) 
structure is the heart of the project.

> Anyway, I'd like to open a discussion about moving in this direction.
> I'd primarily like to hear from John H. and John S. as I suspect that
> this may be pretty disappointing to them.
>
> My thoughts in a bulleted list:
>
>   o Why not Zope?
>
>     There is no denying that Zope has some pretty exciting
>     technologies.  In particular I am still in love with ZODB.
>     Writing Ack! with just ZODB was easy and elegant and really fueled
>     my interest in Zope.  Zope is basically ZODB with a customized
>     version of Medusa (ZServer) plus a truck load of features to
>     remove some of the tedium of web application design.
>     Unfortunately none of these features includes the sort of thing we
>     need and some of the features that are in Zope have been difficult
>     to accommodate at the low level of development necessary to create
>     a framework for what we need.  Although I can see why Traversal is
>     a Good Thing (tm), I/we really don't need it at all for this
>     project and it has totally gotten in the way in part because it
>     doesn't even work correctly.

What is a Traversal when its at home?

>  I am still intrigued by the idea of
>     just wiring ZODB and Medusa (or Twisted) together and calling it
>     Zip or Pope or something.  This would create a nice and simple
>     framework for using persistent python objects in a web application
>     environment... some other time.
>     
Curious to know what sort of system we need to run ZODB. Could we run in 
on a ibiblio account, or is there some problems. Don't know enough about 
Zope to understand which bit made hosting dificult.

>   o Why RDBMS?
>
>     The relational model and existing RDBM systems are mature
>     technologies that are well and widely understood and has excellent
>     documentation.  In some ways this makes it simpler and easier to
>     use than Zope
>
>   o Why not ORM (Object Relational Mapper)?
>
>     I've been turing around in my head how to implement an ORM for
>     this project and in the end I don't really think we want to do
>     this.  My reasoning is summarized above in that it is very
>     difficult to escape the *relationalness* of an RDBMS, so much so
>     that I question wanting to do so.

Curious about this. Do you have an example?
I've never really been convinced about why we really need objects
for the relativly straight forward tasks needed by the web server.
I think objects will come into focus more when we get onto smart clients
where we may need to do more with the data.

>  Also if we choose to emphasize
>     a clean relational dataset cluttering it with objects will make it
>     unfriendly to other tools.
        Rich
--
Plants for a Future: 7000 useful plants
Web:   http://www.pfaf.org/ same as http://www.comp.leeds.ac.uk/pfaf/
Post:  1 Lerryn View, Lerryn, Lostwithiel, Cornwall, PL22 0QJ
Tel:     01208 872 963 / 0845 458 4719  
Email: webmaster@pfaf.org
PFAF electronic mailing list http://groups.yahoo.com/group/pfaf