While well-intentioned, this test keeps causing problems
due to wrong timezone settings in test mode.
It shouldn't completely abort test execution,
since its more of an environment sanity check than a failed test.
Refactored to mark test skipped (regardless of offset, as long as
its greater than 5 seconds). And skipping tests altogether
on SQLite3 with new supportsTimezoneOverride() check.
SapphireTest->setUp() sets the PHP timezone to UTC (see 59547745),
but SQLite doesn't support this for a DB connection.
Since changing it on a global UNIX system level is infeasible,
the tests need to be skipped.
Leave the decision to the phpunit.xml config (via <get> setting),
or to the individual run via "phpunit <folder> '' flush=1".
Flushing takes multiple seconds even on my fast SSD,
which greatly reduces the likelyhood of developers adopting TDD.
Avoids problems when PHPUnit includes this test through autodiscovery. This is the case when running PHPUnit
with a specific path argument, which overrules phpunit.xml.dist
(e.g. "phpunit sapphire/").
Moved some more important stuff up the page,
rewritten upgrading overview to be more task focused,
reduced excessive use of paragraphs to make the text more scannable.
Necessary to have the <get name="db" /> directive
working in phpunit.xml definitions, which in turns allows
us to use GET parameters to switch the database connection
for running automated tests.
When querying DataObjects by a generic parent class (like SiteTree for instance), fields added via $db
set on child classes wouldnt appear.
This is because Object::__construct wasnt called early enough in DataObject::__construct, so
extensions werent initialised when $db was first accessed
See initial idea at http://open.silverstripe.org/ticket/6441. Added $template property and corresponding getters / setters for customizing the template used. Added relevant unit test.
Broke when CMS URL already had query params,
such as ?locale=en_US with the Translatable module enabled.
With this patch it doesn't double-concat ?locale....
Still doubles query params, but that's acceptable
until we find a more solid URL manipulation lib for JS.
If the applyRelation() was passed a relation that went to a class with a parent
class with a database table, applyRelation would return the name of the parent
class, rather than the class the relation was actually too.
As ModelAdmin puts classes directly into links, a namespaced class
has its \s turned into /s, breaking the links. This escapes them by
converting each \ to a -, then converting them back when loading.
Some of the SQLQuery recipes are now handled better by DataList,
others should be contrasted with their respective DataList implementation,
pointing out the limitations