Re: Gnome Web Plone Sprint

[Carsten Senger <senger rehfisch de>]

Murray Cumming schrieb:
> On Tue, 2008-10-28 at 19:38 +0100, Carsten Senger wrote:
> > Hi Murray,
> >
> > Murray Cumming schrieb:
> > > On Mon, 2008-10-27 at 18:15 -0700, Jonathan Wilde wrote:
[...]

> > For quick tests on a local maschine, we could provide a new Data.fs.
>
> I think it is worthwhile to allow it to be tested regularly.
>
> How else would more than one person work on the system anyway, if the
> data is only one one person's hard drive? I guess I'm not understanding
> something.
>
> > But we will have additions and changes regulary that will be persistent 
> > into the Data.fs during the installation. So every Data.fs with a 
> > installed plone-site will be outdated soon. This Path is a waste of time.
>
> Fair enough. You understand the system. But I don't yet understand what
> the correct way is.

The Database (Data.fs) doesn't only contain the content. It contains 
configuration, in some cases data-schemata and first of all python 
objects (The ZODB is and objectdatabase for python objects).

If we change a datastructure in a [e.g. contenttype]-class, the 
instances of this class needs a migration. So in some cases we would 
need to write a migraion if we refactor something during the next weeks. 
Like you need it in an rdbms, if you change the schema.
If we don't do that, an Object in an existing Data.fs is broken.

It's easier and less error-prone to hold an exported version of the 
content etc. in svn and make sure the import-script works, than walk 
through the Db and fix broken objects.

So I meant that we could provide an Data.fs-file in the meantime until 
an installscript is there, but there is no guarante that it isn't 
outdated in two weeks.

..Carsten