Index: openacs-4/packages/acs-core-docs/www/subsites-design.html =================================================================== RCS file: /usr/local/cvsroot/openacs-4/packages/acs-core-docs/www/subsites-design.html,v diff -u -r1.5 -r1.5.2.1 --- openacs-4/packages/acs-core-docs/www/subsites-design.html 7 Mar 2002 06:55:36 -0000 1.5 +++ openacs-4/packages/acs-core-docs/www/subsites-design.html 15 May 2002 23:26:18 -0000 1.5.2.1 @@ -1,109 +1,50 @@ - -
- -*Note* This document has not gone through the any of the
+ *Note* This document has not gone through the any of the
required QA process yet. It is being tagged as stable due to high
-demand. An OpenACS 4 subsite is a managed suite of applications that work together for
+demand. An OpenACS 4 subsite is a managed suite of applications that work together for
a particular user community. This definition covers a very broad range of
requirements: from a Geocities style homepage where a user can install
whatever available application he wants (e.g. a single user could have their
own news and bboard), to a highly structured project subsite with multiple
interdependent applications. Thus, flexibility in application deployment is
-the overarching philosophy of subsites. Meeting such broad requirements of flexibility demands architecture-level
+the overarching philosophy of subsites. Meeting such broad requirements of flexibility demands architecture-level
support, i.e. very low level support from the core OpenACS 4 data model. For
example, the subsites concept demands that any package can have multiple
instances installed at different URLs - entailing support from the APM and
the Request Processor. Since the design and implementation directly
associated with subsites is actually minimal, a discussion of subsites design
is, in fact, a discussion of how core OpenACS 4 components implicitly support
-subsites as a whole. The subsites problem actually has several quite diverse origins. It was
+subsites as a whole. The subsites problem actually has several quite diverse origins. It was
originally recognized as a toolkit feature in the form of
"scoping". The basic concept behind scoping was to allow one scoped
OpenACS installation to behave as multiple unscoped OpenACS installations so that one
OpenACS install could serve multiple communities. Each piece of application data
was tagged with a "scope" consisting of the (user_id, group_id,
scope) triple. In practice the highly denormalized data models that this
method uses produced large amounts of very redundant code and in general made
-it an extremely cumbersome process to "scopify" a module. Before the advent of scoping there were several cases of client projects
+it an extremely cumbersome process to "scopify" a module. Before the advent of scoping there were several cases of client projects
implementing their own version of scoping in special cases. One example being
the wineaccess multi-retailer ecommerce. (Remember the other examples and get
-details. Archnet?, iluvcamp?) The requirements of all these different projects vary greatly, but the one
+details. Archnet?, iluvcamp?) The requirements of all these different projects vary greatly, but the one
consistent theme among all of them is the concept that various areas of the
web site have their own private version of a module. Because this theme is so
dominant, this is the primary problem that the OpenACS4 implementation of
-subsites addresses. The current implementation of package instances and subsites allows
+subsites addresses. The current implementation of package instances and subsites allows
extremely flexible URL configurations. This has the benefit of allowing
multiple instances of the same package to be installed in one subsite, but
can potentially complicate the process of integrating packages with each
other since it is likely people will want packages that live at non standard
URLs to operate together. This requirement would cause some packages to have
more configuration options than normal since hard-coding the URLs would not
-be feasible anymore. This section will cover all the APIs relevant to subsites, and so will
-consist of portions of the APIs of several systems. Packages The following package is provided for instantiation of packages. The
+be feasible anymore. This section will cover all the APIs relevant to subsites, and so will
+consist of portions of the APIs of several systems. Packages The following package is provided for instantiation of packages. The
apm_package.new function can be used to create a package of any type known to
the system. The apm_package_types table can be queried for a list of
-installed packages. (See APM docs for more detail XXX: insert link here) Site Nodes This data model keeps track of what packages are being served from what
+ Site Nodes This data model keeps track of what packages are being served from what
URLs. You can think of this as a kind of rp_register_directory_map on drugs.
This table represents a fully hierarchical site map. The directory_p column
indicates whether or not the node is a leaf node. The pattern_p column
indicates whether an exact match between the request URL and the URL of the
node is required. If pattern_p is true then a match between a request URL and
a site node occurs if any valid prefix of the request URL matches the site
node URL. The object_id column contains the object mounted on the URL
-represented by the node. In most cases this will be a package instance. The following package is provided for creating nodes. The following package is provided for creating nodes. Request Processor Once the above APIs are used to create packages and mount them on a
+ Request Processor Once the above APIs are used to create packages and mount them on a
specific site node, the following request processor APIs can be used to allow
-the package to serve content appropriate to the package instance. The subsites implementation doesn't really have it's own data
model, although it depends heavily on the site-nodes data model, and the APM
-data model. The primary elements of the subsite user interface consist of the subsite
+data model. The primary elements of the subsite user interface consist of the subsite
admin pages. These pages are divided up into two areas: Group administration,
and the site map. The group administration pages allow a subsite
administrator to create and modify groups. The site map pages allow a subsite
administrator to install, remove, configure, and control access to packages.
The site map interface is the primary point of entry for most of the things a
-subsite administrator would want to do. The current subsites implementation addresses the most basic functionality
+subsite administrator would want to do. The current subsites implementation addresses the most basic functionality
required for subsites. It is likely that as developers begin to use the
subsites system for more sophisticated projects, it will become necessary to
develop tools to help build tightly integrated packages. The general area
@@ -297,37 +210,6 @@
a particular configuration of site nodes/packages. As we build more
fundamental applications that can be applied in more general areas, this
feature will become more and more in demand since more problems will be
-solvable by configuration instead of coding.
+installed packages. (See APM docs for more detail XXX: insert link here)
create or replace package apm_package
as
@@ -160,18 +101,15 @@
show errors
-
-
+represented by the node. In most cases this will be a package instance.
create table site_nodes (
node_id constraint site_nodes_node_id_fk
@@ -199,9 +137,7 @@
);
-
-
+
create or replace package site_node
as
@@ -246,47 +182,24 @@
show errors
-
-
+the package to serve content appropriate to the package instance.
[ad_conn node_id]
[ad_conn package_id]
[ad_conn package_url]
-
-