• last updated 17 hours ago
Constraints
Constraints: committers
 
Constraints: files
Constraints: dates
updated tree

added tree_increment_key() function

tree.increment_key() now does the lpadding for us

  1. … 2 more files in changeset.
postgres does not like a certain flavor of table aliases

fixing creation of site_node under postgres

  1. … 2 more files in changeset.
re-ordering queries in acs_permission.permission_p for performance's sake

    • -33
    • +35
    ./oracle/acs-permissions-create.sql
moved portal::mapping api and data model from new-portal into the core, i.e.

into acs-tcl and acs-kernel. renamed it from portal::mapping to

site_node_object_map, since that is what it is. enjoy.

    • -0
    • +65
    ./oracle/site-node-object-map-create.sql
    • -0
    • +10
    ./oracle/site-node-object-map-drop.sql
    • -0
    • +52
    ./postgresql/site-node-object-map-create.sql
    • -0
    • +10
    ./postgresql/site-node-object-map-drop.sql
  1. … 21 more files in changeset.
adding variation on function

Performance enchancements based on our testing

    • -0
    • +17
    ./oracle/upgrade/upgrade-4.2.1-4.2.2.sql
added "(select acs.magic_object_id('registered_users') from dual)"

clause to registered_users view based on results from scalability testing.

added site_nodes_parent_id_idx based on results from scalability testing

moving ben's tree sortkey utilities for oracle from forums to acs-kernel

    • -0
    • +193
    ./oracle/tree-create.sql
    • -0
    • +18
    ./oracle/tree-drop.sql
  1. … 4 more files in changeset.
using connect by in oracle version of update_last_modified (plpgsql version is still using recursive version)

forgot to pass last_modified in recursive call

created acs_object.update_last_modified procedure that takes in an object_id

and an optional timestamp and sets the last_modified time of the object and

all of it's parents (not including object 0) to the last_modified time passed

in or sysdate if none was passed in.

    • -0
    • +28
    ./postgresql/acs-objects-create.sql
fixed bug in apm_package_version__upgrade_p

    • -0
    • +56
    ./postgresql/upgrade/upgrade-4.2-4.5.sql
file upgrade-4.2-4.5.sql was initially added on branch oacs-4-5.

    • -0
    • +0
    ./postgresql/upgrade/upgrade-4.2-4.5.sql
PG-7.2.1 returns false for 0<2^7. strange indeed. 2^7>0 returns true

Merged with the 4.5 branch (oacs-4-5 tag "don-merge-1")

    • -45
    • +8
    ./postgresql/rel-constraints-create.sql
  1. … 58 more files in changeset.

Mostly fixed subsite issues with the PG version.

    • -45
    • +8
    ./postgresql/rel-constraints-create.sql
  1. … 2 more files in changeset.
added parent_id function to apm_package package

adding an index to increase performance of some views

adding an index to make the registered_users view faster

Merged with OpenACS 4.5 beta1 branch.

    • -1
    • +15
    ./postgresql/community-core-create.sql
  1. … 86 more files in changeset.

Yep, previous Oracle fix had a problem! "case" isn't allowed in PL/SQL???

1. Community core change was due to an earlier patch that allows a person

to register as a user.

2. Deleting an acs-object would fail if the object specific table didn't exist.

acs-objects and the content repository shouldn't require type specific tables

be named when you extend a type, this is a temporary hack which acknowledges

that folks have been cheating by name non-existent tables. Since I plan to

remove the restriction anyway, this is not a big deal.

The Oracle version still needs testing ...

    • -1
    • +15
    ./postgresql/community-core-create.sql

1. Changed a lot of package info files to fix problems ...

2. Added Vinod's bboard patch

3. Rationalized party e-mail checking by putting a "before insert or update"

trigger on the table that lower-cases email (this was already done by the

party.new() func but at least one update missed this). Got rid of all

"lower(email) = " and "upper(email) = " checks and just made all use the

db email column directly and lower-cased Tcl vars. In this way the index

on the email column is always and consistently used without needing to

define functional indexes on the column (an EE feature for Oracle).

4. Rationalized naming of a couple of workflows and fixed a glossary PG bug

where apparently its workflow got overwritten by one from the CMS at some

point in time.

5. Made the .cmp/.dat/.frm filters in the template code usable though not

very beautiful nor exactly correct. But with practice you can use them

to debug pages in the packages/* hierarchy. I also turned these off by

default as there's no need to run these filters on a production site!

6. Performed minor surgery on templating, including adding a new

"ad_return_exception_template" that allows you to return a templated

error page and aborts the script.

7. ??? probably but I think this covers most stuff.

  1. … 73 more files in changeset.
don't allow null for supertype in acs_object_type__create_type. default to 'acs_object'

Added a Ben Adida fix from the Oracle version.

fixed order of loading to allow all_object_party_privilege_map to be created before acs_gropu body