65685[TASK] Fix phpstan checkFunctionArgumentTypes errors in ext:core DependencyInjection
Commit message
Review
-
Code review CR +2 Done
- Benni Mack +2
- Oliver Klee +1
- Sascha Egerer +1
- Jörg Bösche +1
-
Verified V +2 Done
- Benni Mack +2
- Sascha Egerer +1
- TYPO3com +1 CI
Issues 1
Files 2
Review log 2427
Show 2 earlier messages
-
Sascha Egerer uploaded patch set 1 09/09/20, 17:17
-
Sascha Egerer wrote 09/09/20, 17:19
Topic set to phpstan
-
Sascha Egerer updated the commit message in patch set 2 09/09/20, 17:19
-
Oliver Klee voted on patch set 2 CR +1 09/09/20, 18:33
-
Sascha Egerer commented on patch set 2 CR -1 09/10/20, 07:06
ContainerBuilder.php:153I would add some logging here before merging this patch.ServiceProviderRegistry.php:21UnresolvedNot required anymore -
Daniel Goerz updated the commit message in patch set 3 09/10/20, 08:48
-
TYPO3com voted on patch set 3 V +1 09/10/20, 09:40 Completed build in 40m 51s on Thu, 10 Sep, 09:40 AM Test Summary: 73173 passed Find logs and detail information at https://bamboo.typo3.com/browse/CORE-GTC-5475
-
Oliver Klee voted on patch set 3 CR +1 09/10/20, 19:18
-
Sascha Egerer uploaded patch set 4 09/10/20, 19:46
-
Sascha Egerer commented on patch set 4 CR +1 09/10/20, 19:47
ContainerBuilder.php:153ReplyDone -
TYPO3com voted on patch set 4 V +1 09/10/20, 19:57 Completed build in 11m 06s on Thu, 10 Sep, 07:57 PM Test Summary: 73172 passed Find logs and detail information at https://bamboo.typo3.com/browse/CORE-GTC-5518
-
Oliver Klee voted on patch set 4 CR +1 09/10/20, 22:26
-
Daniel Goerz voted on patch set 4 CR +1 09/11/20, 07:18
-
Sascha Egerer commented on patch set 5 09/11/20, 09:37
Published edit on patch set 4. -
TYPO3com voted on patch set 5 V +1 09/11/20, 09:49 Completed build in 11m 31s on Fri, 11 Sep, 09:49 AM Test Summary: 73215 passed Find logs and detail information at https://bamboo.typo3.com/browse/CORE-GTC-5532
-
Oliver Klee voted on patch set 5 CR +1 09/11/20, 10:02
-
Sascha Egerer voted on patch set 5 V +1 CR +1 09/11/20, 10:05
-
Jörg Bösche voted on patch set 5 CR +1 09/11/20, 12:17
-
Benni Mack voted on patch set 5 V +2 CR +2 09/12/20, 18:44
-
Benni Mack merged patch set 6 09/12/20, 18:44
-
Benjamin Franzke commented on patch set 6 09/13/20, 04:50
Post merge -1ServiceProviderRegistry.php:119This change is wrong. "Package" was used by intention here instead of PackageInterface. $package->getServiceProvider() as used below is not available on PackageInterface. Please tell me, what was "incompatible" about that type? This looks like a unrelated refactoring that blindly applies the Interface suffix, without checking that all used methodes of that method argmument are actually available on the Interface. This really hurts code quality, if we "Fix" things in a way that breaks stuff by changing method requirements to be actually incorrect. I'm not an IDE user, but I wonder: Isn't such change highlighted in your editors in a way that shows that getServiceProvider() does not exist on PackageInterface? At least phpstan should have shown actually one more error than before with this change (but I guess that is muted). -
Benjamin Franzke commented on patch set 6 09/13/20, 04:59
ServiceProviderRegistry.php:119Reply UnresolvedLet's revert this: https://review.typo3.org/c/Packages/TYPO3.CMS/+/65709 -
Benni Mack commented on patch set 6 09/13/20, 08:57
> Patch Set 6: > > (1 comment) Thanks, you are completely right. I thought this was a good fix, but wasn't aware that we don't have that in the interface. -
Alexander Schnitzler commented on patch set 6 09/14/20, 07:48
> Patch Set 6: > > > Patch Set 6: > > > > (1 comment) > > Thanks, you are completely right. I thought this was a good fix, but wasn't aware that we don't have that in the interface. Sorry to say but using an interface at all makes no sense then. If the package implementation isn't replaceable, the interface has no use and is quite misleading. And on top: If the package isn't really replaceable, we can easily adjust the interface because it's not in use by others out there. So, no breaking change imho. -
Benni Mack commented on patch set 6 09/14/20, 07:50
> Patch Set 6: > > > Patch Set 6: > > > > > Patch Set 6: > > > > > > (1 comment) > > > > Thanks, you are completely right. I thought this was a good fix, but wasn't aware that we don't have that in the interface. > > Sorry to say but using an interface at all makes no sense then. If the package implementation isn't replaceable, the interface has no use and is quite misleading. And on top: If the package isn't really replaceable, we can easily adjust the interface because it's not in use by others out there. So, no breaking change imho. That is because the "Package" concept came from Flow back then, where we had a PackageInterface, and Packages, removing the Interface would make sense but would be breaking, which we never did before. -
Alexander Schnitzler commented on patch set 6 09/14/20, 07:56
> Patch Set 6: > > > Patch Set 6: > > > > > Patch Set 6: > > > > > > > Patch Set 6: > > > > > > > > (1 comment) > > > > > > Thanks, you are completely right. I thought this was a good fix, but wasn't aware that we don't have that in the interface. > > > > Sorry to say but using an interface at all makes no sense then. If the package implementation isn't replaceable, the interface has no use and is quite misleading. And on top: If the package isn't really replaceable, we can easily adjust the interface because it's not in use by others out there. So, no breaking change imho. > > That is because the "Package" concept came from Flow back then, where we had a PackageInterface, and Packages, removing the Interface would make sense but would be breaking, which we never did before. If noone can use the interface because it does not define api, then changing it isn't breaking. What's the reason to use it other than in xclasses, which we don't define as potentially breaking either? -
Sascha Egerer commented on patch set 6 09/14/20, 12:56
> Patch Set 6: > > > Patch Set 6: > > > > > Patch Set 6: > > > > > > > Patch Set 6: > > > > > > > > > Patch Set 6: > > > > > > > > > > (1 comment) > > > > > > > > Thanks, you are completely right. I thought this was a good fix, but wasn't aware that we don't have that in the interface. > > > > > > Sorry to say but using an interface at all makes no sense then. If the package implementation isn't replaceable, the interface has no use and is quite misleading. And on top: If the package isn't really replaceable, we can easily adjust the interface because it's not in use by others out there. So, no breaking change imho. > > > > That is because the "Package" concept came from Flow back then, where we had a PackageInterface, and Packages, removing the Interface would make sense but would be breaking, which we never did before. > > If noone can use the interface because it does not define api, then changing it isn't breaking. What's the reason to use it other than in xclasses, which we don't define as potentially breaking either? By the way the PackageManager::getPackage() method, which is used in line 125 to overwrite $package, does also return PackageInterface. So that was also "broken" before. IMO its correct to use the interface and fix the interface instead if we do require the method anyways.