TYPO3 Core

The Extension Manger shall not give always an XCLASS warning message to a PHP file. Some cases exist where a XCLASS is not usefull.

Steffen Kamper: "this should be an em_conf flag like
excludeDirsFromChecking => 'res/3rdparty,res/test' "

Ernesto Baschny has written on the Core list under topic "RFC #23396: remove code warnings about XCLASS and naming ":

The problem is if you have an extension tree, you cannot XCLASS a base
class. It doesn't make sense and it doesn't work. This isn't only
extbase related.

E.g.:

class tx_myext_person
    class tx_myext_boss extends tx_myext_person

Now imagine a XCLASS in tx_myext_person (to keep current EM happy):

In my code I do makeInstance(tx_myext_boss) and never
makeInstance(tx_myext_person) (maybe it is even an abstract base class).
So there is no way of XCLASSing tx_myext_person and so no reason to have
XCLASS information added to that file.

So I agree that the EM XCLASS warnings have to either go away or its
code optimized and moved over to some helper module of extdeveval. But I
would opt for only removing it from core as soon as we have a
replacement in extdeveval. Else we end up with no check at all, which
would be sad.

So until then a -1 on the current patch, but a +1 on the general idea.
We need to get extdeveval team involved and I'll try to make the
necessary connections here. Olly Hader seems the "most active" developer
on the extdeveval front. Franz, do you mind trying to create a patch for
current extdeveval trunk to include that checks somewhere? Opening an
issue on forge for that (with patch) would sure speed up the process. [1]

Even better:

'excludeXCLASScheck' => array(
'res/',
'special/abc*.php',
),

+1 to that. Please come up with a new RFC for that, it someone manages
to create that patch. Thanks!

Cheers,
Ernesto

(issue imported from #M16178)