drupal_render(): invert second argument ($is_recursive_call -> $is_root_call) => more strict, better DX/TX

#2331393: drupal_render() recursive parameter usage is unclear, safe default should be drupal_render(x) was the original issue.

Lets close one as a duplicate, please.

Thanks for your work!

1. +++ b/core/authorize.php
   @@ -104,7 +104,7 @@ function authorize_access_allowed() {
   -    $output = drupal_render($authorize_report);
   +    $output = drupal_render($authorize_report, TRUE);
   

   I updated the IS to make clear that drupal_render() is the common case for contrib ...

2. +++ b/core/includes/common.inc
   @@ -2708,18 +2708,27 @@ function drupal_prepare_page($page) {
   - * @param bool $is_recursive_call
   + * @param bool $is_root_call
     *   Whether this is a recursive call or not, for internal use.
     *
     * @return string
   

   Can you please update the documentation to clearly define when to pass TRUE and when not? You should not have to read the entire drupal_render() documentation for it.

3. +++ b/core/includes/common.inc
   @@ -2708,18 +2708,27 @@ function drupal_prepare_page($page) {
     */
   -function drupal_render(&$elements, $is_recursive_call = FALSE) {
   +function drupal_render(&$elements, $is_root_call = FALSE) {
   

   ... just as a follow up, I think it would be nice to have a drupal_render_root() which just does drupal_render(..., TRUE) which would be a little bit easier to read in the code.

4. +++ b/core/modules/ckeditor/ckeditor.module
   @@ -96,3 +96,75 @@ function _ckeditor_theme_css($theme = NULL) {
   + */
   +function ckeditor_rebuild() {
   +  /** @var \Drupal\filter\FilterFormatInterface[] $formats */
   +  $formats = filter_formats();
   

   Did you considered to have a wrapper around the cache entry instead and just invalidate it in hook_rebuild()?

5. +++ b/core/modules/views/src/Plugin/views/area/HTTPStatusCode.php
   @@ -64,8 +64,8 @@ public function buildOptionsForm(&$form, FormStateInterface $form_state) {
        if (!$empty || !empty($this->options['empty'])) {
   -      $this->view->getResponse()->setStatusCode($this->options['status_code']);
   -      $this->view->getRequest()->attributes->set('_http_statuscode', $this->options['status_code']);
   +      $build['#attached']['http_header'][] = ['Status', $this->options['status_code']];
   +      return $build;
        }
   

   Let's remove that hunk for now.

+++ b/core/authorize.php
@@ -104,7 +104,7 @@ function authorize_access_allowed() {
-    $output = drupal_render($authorize_report, TRUE);
+    $output = drupal_render_root($authorize_report);

Nice, this is much more explicit

RTBC + 1, but needs a short change record - if I see this correctly - or rather update the existing one.

Leaving as RTBC for core committers to decide what needs to be documented here as change.

-
+    $format_tags = $this->cache->get('ckeditor_internal_format_tags:' . $format->id())->data;
     return $format_tags;

This has no fallback for if the cache item goes missing at all - for example a memcache eviction.

Should it go into state instead?

RTBC + 1 again

FWIW, I saw it and wondered why there was no rebuild called on cache miss, but figured there some something else I was missing and given there was a rebuild function added I figured it would be called in that case.

Next time will speak up. Thanks!

Speaking of that:

- What happens if state is not yet populated at that point?

Yes that's a good question, needs a default to empty array in the case of no formats saved yet I think?

Back to RTBC then :).

I asked Wim why we still even call check_markup() at all, and he reminded me it's been left around only for the case where you only want the final rendered string (and to completely discard #attached etc.), for example in an e-mail.

This means that https://api.drupal.org/api/drupal/core%21modules%21views%21src%21Plugin%... is using it wrong.

But ckeditor, while it's doing strange things here, is using it right.

I didn't have any other feedback apart from #12, and that's been dealt with.

So... Committed/pushed to 8.0.x, thanks!