User-Agent: Microsoft-Entourage/11.2.1.051004
Date: Tue, 06 Dec 2005 22:50:19 +0000
Subject: Re: Integration Requirements
From: Dave Page <dpage@vale-housing.co.uk>
To: Josh Berkus <josh@agliodbs.com>,
	<pgsql-www@postgresql.org>
Message-ID: <BFBBC62B.737D%dpage@vale-housing.co.uk>
Thread-Topic: [pgsql-www] Integration Requirements
Thread-Index: AcX6t26DrN3RZmaqEdq9WAARJHpWaA==
In-Reply-To: <200512060924.35627.josh@agliodbs.com>
Mime-version: 1.0
Content-type: text/plain;
	charset="US-ASCII"
Content-transfer-encoding: 7bit


On 6/12/05 5:24 pm, "Josh Berkus" <josh@agliodbs.com> wrote:

> Dave,
> 
> The below go far beyond "integration requirements" and stray into the area of
> dictating exactly how new web code should be built.  If you do that, you
> won't get any contributors.  Would you write WWW code on your own time if
> someone told you exactly what programming structures to use?

I did. The last website. I am not dictating anything however - my comments
are a summary of what seemed to me to be the final agreement on what was
going to be done last time round. If you've got better ideas, let's hear
'em.

> Integration requirements would be something like:
> -- Needs to be able to run on FreeBSD
> -- Must use the postgresql.org CSS, which is documented in __________ (note
> Magnus' comment on the lack of CSS documentation).
> -- Can't have processor/ram requirements beyond ______________ unless a server
> is going to be donated,
> -- Must be in one of the following programming languages: PHP, Perl, Ruby ...
> if in something other than PHP or Perl must be part of a long-term commitment
> for code maintenance.
> etc.
 
- Must be written in PHP4, website template format, or a mix of the two.
- Must use the postgresql.org css if dynamically rendering pages in PHP.
- Must use 'friendly' URLs without POST or GET variables controlling
content, to allow mirroring (friendly URLs can be rewritten in GET
variables, per existing pages on the site).

>> - Moving data; originally we'd looked at exporting from the cms into the
>> filesystem, and having script that did a cvs add/remove/commit over the
>> entire tree, into the main web CVS. This is still preferrable from an 'ease
>> of rebuilding' point of view, but might be easier just to rsync the content
>> from the filesystem of the cms machine to wwwmaster.
> 
> This isn't really practical for a KB or many other components, which need to
> be highly dynamic, not a bunch of static pages.

It's going to be static if it's on the main website because the site is 100%
static. That doesn't mean that page updates cannot be reflected within 15
minutes or an hour or whatever.
 
>> - Navigation; Gevik was working on a tree-style thingy in PHP. Perhaps the
>> CMS could export an XML file describing the navigation tree, which the PHP
>> handler on the main website could use to generate it's treeview in dynamic
>> mode. By accepting some sort of pointer to the currently selected node as a
>> GET value, we should be able to make the tree fully mirrorable.
> 
> This is functional specification based on a lot of assumptions about the shape
> of the final interface, and I can't imagine it even being applicable to
> something I, personally, would build.

This is the design that Gevik proposed, discussed on -www (and -hackers
iirc) and previewed previously. I merely suggested a way for a CMS to
produce output to control the interface. Again, if you have other
suggestions about user interface and how that may be controlled, please
share them.

Unless someone actually proposes new ideas, I can only suggest how the
previous ones might be integrated with the website.

Regards, Dave