Connor

During porting, I change the `is` method to the more type-strict `assertSame` method, so I look at the tester's intent in each case. However, here the tester intent is wrong.

Another "who made a mistake? / where is the bug?" memo `sfWebRequestTest.php` ```php $request->languages = null; $_SERVER['HTTP_ACCEPT_LANGUAGE'] = ''; $this->is($request->getLanguages(), [], '->getLanguages() returns an empty array if the client send...

Another "who made a mistake? / where is the bug?" memo `sfI18NTest.php` ``` $t->is($i18n->getTimeForCulture(null), null, '->getTimeForCulture() returns null in case of conversion problem'); ``` Test message says: `->getTimeForCulture() returns null...

sfViewCacheManager::has() method with null return https://github.com/FriendsOfSymfony1/symfony1/blob/2826410bbfdb8dd69db968b9c44ca2550a38a9da/lib/view/sfViewCacheManager.class.php#L384-L391

> [#286 (comment)](https://github.com/FriendsOfSymfony1/symfony1/issues/286#issuecomment-1857974108) > > ``` > $acceptLanguage = '0'; > empty($acceptLanguage) // true > ``` > > Does a string "0" is an empty ACCEPT_LANGUAGE header ? An empty...

Side note: we should deprecate and totally remove pear support (plugin component). PEAR is a deprecation hell and not working with composer.

A skeleton jumped out of the closet. The problem is that the current codebase doesn't recognize and use the composer autoloader, everyone has to load it themselves (see migration wiki).

> I am not sure. > > Indeed adding new dependency that requires the use of composer is a BC break. > > As composer is not yet required. Yes,...

@mentalstring It does not seem very complicated. My biggest dilemma is how to do the separation: actually, or in a subtree layout, as the symfony does. I have similar problems...

This feature protects the developer during development from doing the inappropriate thing (e.g. calling cli.php in the wrong way). But on a complete system, this method is no longer relevant...