Attached patch removes or converts the remaining, trivial usages of variable_get(), variable_set(), and variable_del():
https://api.drupal.org/api/drupal/core!includes!bootstrap.inc/function/c...
https://api.drupal.org/api/drupal/core!includes!bootstrap.inc/function/c...
https://api.drupal.org/api/drupal/core!includes!bootstrap.inc/function/c...
Without these, the only two major usages are the language system and node/search rankings:
#1862202: Objectify the language system
blocks #2108599: Convert language_default to CMI
blocks #2102477: Convert remainder of language negotiation settings to configuration system
#2042807: Convert search plugins to use a ConfigEntity and a PluginBag
Comment | File | Size | Author |
---|---|---|---|
#3 | drupal8.variable-leftovers.2.patch | 3.77 KB | sun |
drupal8.variable-leftovers.0.patch | 5.48 KB | sun | |
Comments
Comment #1
alexpottSurely we can just remove the variable_set() nothing is using variables that are set and the tests are passing :)
This change is already being attempted in #2108679: API documentation: Convert my_module_menus examples in menu.api.php to CMI how about reviewing that?
Comment #2
sunComment #3
sunre: hook_menu_*()
ACK, followed up + provided new patch over there. Thanks for the pointer!
(OT: I currently have the Drupal core dev experience that 99% of the rest of the world probably has, too... there are so many issues, each one covering super-small bits only, it's impossible to make sense of what's going on [and what's not] → "Issueitis" ;))
re: RDF test:
Yes, the lack of that node type settings update should technically cause the test to fail, and because it apparently does not, we can only assume that the manipulation is not required in the first place. Happy to simply kill those lines.
re: RemoteFileUnmanagedMoveTest:
This one, however, is different — the lack of setting that override merely means that the (actually executed) parent
UnmanagedMoveTest
is essentially executed twice without actually testing the case of remote files/schemes/streams.By putting the override back in place (copied 1:1 from other Remote* tests), this test actually re-starts to test what it is supposed to test. The fact that tests are still passing without and with that override just means that the functionality still works as expected. — Phew, lucky us! :)
Comment #4
alexpottLooks good - thanks.
Comment #5
webchickAwesome!
Committed and pushed to 8.x. Thanks!