• last updated 23 hours ago
Constraints
Constraints: committers
 
Constraints: files
Constraints: dates
1. Added "--default foo" style comments to new PL/pgSQL function parameters

because xotcl-core, rather than use the metadata created by define_function_params,

ASSIGNS SEMANTICS TO THESE SQL COMMENTS. Bah. If I truly cared about quality

I'd insist the xotcl people fix their code and do things right, but I know that

would never happen so ... kludge away.

2. xotcl-core adds CR attributes (rather than fix core). So my fixing of

core conflicts, meaning I have to kludge again. Still needs to be fixed

for oracle.

    • -0
    • +197
    ./postgresql/upgrade/upgrade-5.7.0d9-upgrade-5.7.0d10.sql
  1. … 2 more files in changeset.
Added upgrade script to redefine a couple of plpgsql functions that fail with

later versions of PG due to tighter type checking.

    • -0
    • +78
    ./postgresql/upgrade/upgrade-5.7.0d8-5.7.0d9.sql
  1. … 1 more file in changeset.
Fixing many bugs

    • -0
    • +191
    ./postgresql/upgrade/upgrade-5.7.0d7-5.7.0d8.sql
    • -67
    • +0
    ./postgresql/upgrade/upgrade-5.7.0d7-upgrade-5.7.0d8.sql
Hopefully got the friggin' oracle stuff to work, not to mention finding

pg kernel bugs and acs-object-management bugs.

    • -46
    • +326
    ./oracle/acs-metadata-create.sql
    • -507
    • +0
    ./oracle/upgrade/upgrade-5.7.0d4-5.7.0d5.sql
Had to remove the cascade_p feature of the new drop_type procedure, because

implementing it in Oracle would require a huge reorganization of the order

of definition of object and metadata tables and their associated tables,

which just is not going to happen.

    • -19
    • +5
    ./postgresql/acs-metadata-create.sql
    • -0
    • +67
    ./postgresql/upgrade/upgrade-5.7.0d7-upgrade-5.7.0d8.sql
  1. … 1 more file in changeset.
Final version of the 5.7 oracle upgrade for acs-kernel (unless CR work

uncovers problems).

    • -13
    • +61
    ./oracle/upgrade/upgrade-5.7.0d6-5.7.0d7.sql
Intermediate check-in of kernel changes for 5.7

    • -0
    • +647
    ./oracle/upgrade/upgrade-5.7.0d6-5.7.0d7.sql
*** empty log message ***

Further bug fixing ... hopefully, testing with AOM will tell the tale.

Also bumped *info file.

  1. … 1 more file in changeset.
Fixed a bug with the new version of create_type, found as I was porting to

Oracle (there *is* a use for Oracle, I'll be damned!).

    • -0
    • +117
    ./postgresql/upgrade/upgrade-5.7.0d5-5.7.0d6.sql
Believe it or not, "create_type" has never, ever checked to see if the

given supertype is a valid type. The code now verifies that the given

supertype is a child of "acs_object", except when supertype is passed in

as NULL or '' (in which case it's set to acs_object). Only did it for

PostgreSQL since it's been around forever and whatever Oracle users don't

know, won't hurt them.

    • -0
    • +119
    ./postgresql/upgrade/upgrade-5.7.0d4-5.6.0d5.sql
  1. … 1 more file in changeset.
First cut at Oracle upgrade for enhanced acs_datatypes and create_type,

create_attributes functionality that's already been completed for PG.

DO NOT ATTEMPT TO USE, IT'S NOT FINISHED.

    • -0
    • +507
    ./oracle/upgrade/upgrade-5.7.0d4-5.7.0d5.sql
Moved table building and column adding code from content repository to

kernel, so all objects can benefit. Part of the acs object management package

work.

    • -159
    • +493
    ./postgresql/acs-metadata-create.sql
    • -309
    • +5
    ./postgresql/upgrade/upgrade-5.7.0d3-5.7.0d4.sql
  1. … 4 more files in changeset.
Upgrade scripts for changes on object metadata creation (adding automatic

table and column creation to base objects as well as CR object, object views,

etc).

    • -0
    • +936
    ./postgresql/upgrade/upgrade-5.7.0d3-5.7.0d4.sql
  1. … 3 more files in changeset.
On my previous commit I added a not null constraint for name column on site_nodes, the problem is that main site node on oracle has a NULL name ( given the fact that empty strings are managed as NULL on Oracle ) therefore doesnt make sense to have such constraint.

- Avoiding usage of coalesce function on site_nodes table columns in WHERE clause, this was leading to usage of sequencial scans which can be expensive when having a huge amount of site_nodes. Instead we go for isolation of the case when requestion a node with a null parent ( this would be the main site node ) and we use the = operator so the planner goes for a index scan.

- Adding not null constraint to site_nodes(name)

    • -0
    • +9
    ./oracle/upgrade/upgrade-5.7.0d2-5.7.0d3.sql
    • -0
    • +82
    ./postgresql/upgrade/upgrade-5.7.0d2-5.7.0d3.sql
  1. … 1 more file in changeset.
apm_parameter__get_value bug missed by merge apparently.

Merged oacs-5-6 to HEAD. HEAD now installs and passes all core tests. There

may still be some lingering problems due to my screwing up the oacs-5-6

branch earlier, I'll be looking into this later.

    • -0
    • +19
    ./postgresql/upgrade/upgrade-5.6.0b1-5.6.0b2.sql
    • -1
    • +28
    ./postgresql/upgrade/upgrade-5.6.0d1-5.6.0d2.sql
  1. … 207 more files in changeset.
Changes for supporting postgresql 9.0:

- Avoiding usage of SQL reserved keywords on plpsql functions (plpsql lexer and parser were reworked).

- Fixing all those places on which FROM clause was automatically added by PG ( PG Config add_missing_from parameter was removed for this release of PG ).

Many thanks to Tracy Adams and Jeff Lu for providing first version of the patches.

    • -0
    • +103
    ./postgresql/upgrade/upgrade-5.7.0d1-5.7.0d2.sql
  1. … 15 more files in changeset.
file upgrade-5.6.0b2-5.6.0b3.sql was initially added on branch oacs-5-6.

    • -0
    • +0
    ./postgresql/upgrade/upgrade-5.6.0b2-5.6.0b3.sql
Bumped to beta 3 so I could put in this important upgrade script, fixing

a problem with using package_instantiate_object to instantiate a base

(acs_object) object. Never done in core or .lrn code apparently because typically

you instantiate subtypes, but it is important to get right.

  1. … 26 more files in changeset.
Somebody hosed this (probably me). Never caught because no one creates a

base object directly, just subtypes of acs_object with their own (correct)

definition of a new() function.

I somehow managed to avoid commiting the "is_child" addition to apm-create

and also the upgrade script.

- complete change to acs_objects_package_idx for new creates (see http://fisheye.openacs.org/changelog/OpenACS/?cs=oacs-5-6%3Agustafn%3A20100509112929)

Needed to drop bio_mime_type attribute values, too ... this is removed

as it was never actually used so no need to copy it into the persons

table.

Had to remove the drop/create of cc_users because some obscure views make

use of it. No existing code expects cc_users to contain "bio" anyway, so

recreating the view was a matter of tidiness, not need. Drop ... cascade

would work but would break ecommerce and dotlrn-ecommerce, and likely some

other packages, so this seems the lesser of two evils.

    • -14
    • +5
    ./postgresql/upgrade/upgrade-5.6.0b1-5.6.0b2.sql
    • -14
    • +0
    ./postgresql/upgrade/upgrade-5.6.0d1-5.6.0d2.sql
Fixed up some "bio" related upgrade stuff, including protection against

trying to upgrade sites that had already applied the upgrade against 5.5

(that script didn't make it into the repository however).

    • -1
    • +27
    ./postgresql/upgrade/upgrade-5.6.0b1-5.6.0b2.sql
    • -1
    • +33
    ./postgresql/upgrade/upgrade-5.6.0d1-5.6.0d2.sql
- create index for acs_objects on package_ids and drop old one (as discussed in OCT)

- bump version number to trigger run of upgrade scripts for testing

  1. … 1 more file in changeset.
file upgrade-5.6.0b1-5.6.0b2.sql was initially added on branch oacs-5-6.

    • -0
    • +0
    ./oracle/upgrade/upgrade-5.6.0b1-5.6.0b2.sql
    • -0
    • +0
    ./postgresql/upgrade/upgrade-5.6.0b1-5.6.0b2.sql
Making global/instance changes work with Oracle

  1. … 1 more file in changeset.