Sometimes, in order to provide a cross-db behaviour of unique indexes
mixing null and not null columns, we create, under the hood, some
function based indexes.
When that happens get_indexes() is returning the name of the
expression objects used to calculate that function index. But we need
the original column names to be able to compare indexes and get
column dependencies properly.
So, this patch just looks, when the index is unique and function based,
to the expressions (pretty standard CASE statements) and gets the
original column name from it.
Covered with tests.
The PHP_CodeSniffer @codingStandardsIgnore annotations are deprecated
and, since version 3.x, the new // phpcs:ignore comments should be used
instead.
This commits just reviews all the uses in core, replacing them for
the better new candidate, or removing when no longer needed.
Prior to this patch the debugging mode (when enabled) would trigger
on everywhere, regardless of whether or not READ_ONLY_SESSION is defined.
This patch modifies that behaviour so that the debugging only kicks in
if READ_ONLY_SESSION is defined and set to true.
moodle_read_slave_trait: when creating another handle, restore temptables
property that is clobbered by raw_connect().
Also a better condition for temptable related queries detection in
pgsql_native_moodle_database.
dml_pgsql_read_slave_test::test_temp_table(): use real db connection
if possible, otherwise skip the test.
1. pg_connection_status(false) fails on PHP8 because the argument is not a stream
2. skipeed the unittest that expected that pg_query() does not show an error if the resource is invalid
see also MDL-71482
In PHPUnit 9.1, the following regexp-related assertions
have been deprecated and there are new alternatives for
all them:
- assertRegExp() -> assertMatchesRegularExpression()
- assertNotRegExp() -> assertDoesNotMatchRegularExpression()
This is about to, simply, move all cases to the new alternatives.
Source: https://github.com/sebastianbergmann/phpunit/blob/9.1.0/ChangeLog-9.1.md
Regexp to find all them:
ag 'assertRegExp|assertNotRegExp' -li
This patch adds a solution for Postgres that allows overriding IN
statement limit of 16bit (65535 values) by wrapping items in VALUES
list. This only takes place in number of items exceeds 65535, otherwise
parent class get_in_or_equal() function is used.
With MariaDB 10.2.7 the driver was incorrectly returning '' (empty
string) as default for all the columns having null (meaning, no
default).
Also, cover the case in unit tests, it seems that we were already
testing other defaults but not the null case.
The optional parameters of assertEquals() and assertNotEquals()
are deprecated in PHPUnit 8 (to be removed in PHPUnit 9):
- delta => use assertEqualsWithDelta()
- canonicalize => use assertEqualsCanonicalizing()
- ignoreCase => use assertEqualsIgnoringCase
- maxDepth => removed without replacement.
More info @ https://github.com/sebastianbergmann/phpunit/issues/3341
Initial search done with:
ag 'assert(Not)?Equals\(.*,.*,' --php
Then, running tests and fixing remaining cases.
Both assertContains() and assertNotContains() are deprecated in PHPUnit 8
for operations on strings. Also the optional case parameter is. All uses
must be changed to one of:
- assertStringContainsString()
- assertStringContainsStringIgnoringCase()
- assertStringNotContainsString()
- assertStringNotContainsStringIgnoringCase()
More info: https://github.com/sebastianbergmann/phpunit/issues/3422
Regexp to find all uses:
ag 'assert(Not)?Contains\('
While this is not strictly required, because removal will
happen in PHPUnit 9.0, we are already getting rid of all
uses in core.
From release notes:https://phpunit.de/announcements/phpunit-8.html
assertInternalType() is deprecated and will be removed in
PHPUnit 9. Refactor your test to use assertIsArray(), assertIsBool(),
assertIsFloat(), assertIsInt(), assertIsNumeric(), assertIsObject(),
assertIsResource(), assertIsString(), assertIsScalar(),
assertIsCallable(), or assertIsIterable() instead.
All the setup/teardown/pre/post/conditions template methods
now are required to return void. This was warned with phpunit 7
and now is enforced.
At the same time, fix a few wrong function names,
provider data and param types, return statements...
The new API works on normal databases (by deleting data based on the
subquery) and also on MySQL (by deleting the data using a weird join
on the subquery).
Implemented with moodle_read_slave_trait
Functionality is triggered by supplying config dboption['readonly'].
See config-dist.php for more info on supported dboptions.
pgsql and mysqli drivers are using this feature. Also added support for
connection timeout for these two drivers.
If there is a PHP fatal error and destructors do not run (this can
happen in out-of-memory errors, and maybe if there is an error in a
previous destructor) then Postgres cursors may be left open.
Usually this does not cause a problem because the connection is
closed anyway, but if using persistent connections, a future
request may reuse the connection with a cursor open. It then gets
errors when it tries to create a new cursor with the same name.
This change closes all cursors at the start of a persistent
connection.
MySQL v8 has added "groups" to the reserved word list. That is used
as a table name in Moodle so we have to quote it with backticks in the
SQL so that MySQL knows it's being used as an identifier.
It's more memory efficient to use `pg_fetch_assoc` for each row than to
call `pg_fetch_all` and release memory immediately. This is because we
can treat the assoc fetch like an iterator and it only fetches the
current record into memory one at a time, whilst the all fetch fetches
all records and never unsets them. Attempting to unset them is extremely
time consuming.
This is a followup of 85f47ba, where we were relaxing
the (new since phpunit 7.x) strict (===) isEqual()
comparison for strings. Copying the explanations for
easier understanding.
Link: https://github.com/sebastianbergmann/phpunit/issues/3185
Solution here is one of:
a) Return to the previous situation, making the comparison
softer. That can achieved by forcing different types, so
float == string works.
b) Changing APIs (both forms and database return strings) to
perform some conversion to floats. That would make float
comparison (with floats or strings) to work too.
The patch here follows the a) approach. Changing all the internals
for proper float handling sounds excesive when it has been working
perfectly since ever. So we went the easier route, just getting
rid of the new === comparisons when needed by changing expectation
types to float.
