• last updated 12 hours ago
Constraints
Constraints: committers
 
Constraints: files
Constraints: dates
Fix documentation (complete list of array fields)

use separate urlspace for site-node cache to reduce lock contention

Replace removed query in xotcl implementation as well

Init cache using values from db instead than cached api (fixes install from scratch)

  1. … 1 more file in changeset.
- fix caching bug, when urls are retrieved via object_ids, and packages are unmounted

- make code more robust

- extend regression tests

  1. … 1 more file in changeset.
fix the value which is used for the partition_key during flush_cache operation

improve comment

remove obsolete comment

improve Oracle compatibility

last part of site_nodes changes with tree_sortkey dependency

Introduce has_children for SiteNodeUrlspaceCache to avoid queries returning potentially a huge list of results

fix potential problem when updating cache on top-level site

  1. … 1 more file in changeset.
improve comments

  1. … 10 more files in changeset.
Fix typos

  1. … 5 more files in changeset.
fix cache_key name for node_id in flush_cache

scalability improvement: don't force a conversion of the string to a list (for sites, having e.g. several 10k sub nodes on a site node)

on one site, this change reduces the overall processing time of a requests from ~16-20ms to ~4ms

improve spelling

  1. … 6 more files in changeset.
improve log message

Fix parameter names in proc doc

Replace array idioms with dict idioms

flush per-request cache (fixes regression test properly)

Revert change... more investigation needed

node_id shoudl probably be parent_id (fixes automated tests)

improve rowcount expressions

  1. … 76 more files in changeset.
document public methods

it seems, we can reactivate per-request caching for sitenodes, otherwise we see up to 37 locks per requests on get_node_id on openacs.org (regression test works fine)

fix site-node test in regression test

don't use per-request caching, since we migh shoot in our own foot

interrupt cache_transaction to achieve desired semantics

allow in a few cases empty node_ids for method "get_url" as it is used by current OpenACS

  1. … 1 more file in changeset.
Get parent only once

Fix proc and contract doc elements, so they are properly parsed by apidoc::api_*_documentation.

- Add mising @param

- Add missing @author

- @creation_date -> @creation-date

- @creation-user -> @author

- Fix typos

  1. … 8 more files in changeset.
strengthen type checking for site_node ids to make it clear that empty strings are not accepted