will work and where it will not.
This test failed due to an API change in WebGUI. Before 7.3, if you created
a Storage object and then created an Asset with that Storage Id, then WebGUI
would do that happily. This is actually a bug, because you could create multiple
Assets with that one Storage Id, and when _any_ of the Assets was deleted or
rolled back it would delete the Storage location and leave the other Assets
with that Id pointing to an empty Storage object.
So the API change is actually a bug fix, albeit being a side-effect of the
auto-commit work for posts and wikis.
Added a baseline test for WebGUI::International which also tests the new methods.
Fixed codespace caching in WebGUI::Operation::Help. Codespace caching is
done via symbol table lookups, and it saves doing an eval.
Implemented the same type of caching in WebGUI::International. This replaced
the in-memory cache which would slowly accumulate a complete copy of the
i18n entries as they were fetched.
Broke out code scanning tests into their own tests, and required the CODE_COP environment
variable be set before they were run.
Fixed two bugs in the Workflow test, wrong number of tests and inverse test logic.
More Auth template inheritance work. Almost time to reflect it over to
the LDAP module.
It didn't work with CIDR format IP addresses. I cannabilized the
code from canShowDebug and refactored it into its own method,
canShowBasedOnIP. Both canShowDebug and canShowPerformanceIndicators call this.
Updated the docs for the debugIp setting to note that it's used for both
showDebug and showPerformanceIndicators.
Added about 20 more tests to ErrorHandler.t. Only the two hard methods are
left, fatal (which closes the session at the end) and showDebug, which is
validating a ton of output.
Added pluggable template parser docs.
From the dev list, changed HTMLTemplateExpr to translate all dots in template
variables to underscores, since EXPR's don't do dots, just like TemplateToolkit.
More tests for Session/Icon.pm. Due to the way the code is written,
this test will not test 100% of the code. Untestable conditions include
not being able to find the toolbar directory in getToolbarOptions, and
all code that sets a default pageUrl in any icon-specific method.
Documented the disabled flag for moveDown and moveUp. Only the EMS uses this in
the core as far as I can tell.
Optimized out 1 profile call in getBaseURL
There are still a few functional tests that need to be written, they're
codified as TODOs.
In Session/Style.pm, in all methods that have a send immediate option
(when sent=1), they no longer send and cache. This prevents duplicate
heag tags from being generated if generateAdditionalHeadTags was
ever called twice.
fixed a bug in userStyle where it would not process data that was false.
process will do this, so I made userStyle do it as well by adding
in a definedness check on $output.
