[PHPUnitBridge] Improved deprecations display #35271

greg0ire · 2020-01-08T20:55:25Z

Q	A
Branch?	master for features
Bug fix?	no
New feature?	yes
Deprecations?	no
Tickets	n/a
License	MIT
Doc PR	todo

Ignore verbose when the build fails because of deprecations
Add per-group verbosity : it is now possible to make the report a lot quieter, by specifying which groups should become quiet, like this: quiet[]=indirect&quiet[]=direct

Might add more improvements to that branch if people suggest some.

OskarStark

👍🏻😃

ostrolucky

Yay, continuous refactoring, a.k.a leaving the place better than before. I do the same. This bridge especially needs it. Few more like this in future should hopefully come.

src/Symfony/Bridge/PhpUnit/DeprecationErrorHandler/Configuration.php

src/Symfony/Bridge/PhpUnit/DeprecationErrorHandler.php

src/Symfony/Bridge/PhpUnit/DeprecationErrorHandler/DeprecationGroup.php

src/Symfony/Bridge/PhpUnit/DeprecationErrorHandler/DeprecationNotice.php

ostrolucky · 2020-01-16T20:22:34Z

src/Symfony/Bridge/PhpUnit/DeprecationErrorHandler/DeprecationNotice.php

+        ++$this->count;
+    }
+
+    public function countsByCaller()


Suggested change

public function countsByCaller()

public function getMethodOccurences()

The method where the deprecation is triggered is not to be confused with the method that calls that method, and countsByCallingMethod seems a bit long, WDYT?

I don't know, they seem symmetric to me so I would expect similar name. I guess we need 3rd opinion.

getCountsByCaller?

src/Symfony/Bridge/PhpUnit/DeprecationErrorHandler.php

greg0ire · 2020-01-20T21:55:55Z

I added per-group verbosity, the usage is verbose[direct]=0|1, but maybe I should use verbose[]=direct, tell me what you think.

src/Symfony/Bridge/PhpUnit/DeprecationErrorHandler/Configuration.php

src/Symfony/Bridge/PhpUnit/Tests/DeprecationErrorHandler/partially_quiet.phpt

src/Symfony/Bridge/PhpUnit/DeprecationErrorHandler/Configuration.php

ostrolucky · 2020-01-20T22:43:50Z

usage is verbose[direct]=0|1, but maybe I should use verbose[]=direct, tell me what you think.

[key]=int is more consistent with what we have now

greg0ire · 2020-01-26T11:50:56Z

I just pushed a version with a test case that makes more sense. Also, I added a commit to remove the "remaining" prefix

nicolas-grekas

Thanks, here are my comments.
Can you please add a line in the changelog of the bridge?
Can you please also update to the description of the PR so that it documents what this is about? I had to review in details to get it, the ppl that will write the doc and/or the blog post about this need help too.

src/Symfony/Bridge/PhpUnit/DeprecationErrorHandler.php

src/Symfony/Bridge/PhpUnit/DeprecationErrorHandler/Configuration.php

src/Symfony/Bridge/PhpUnit/DeprecationErrorHandler/DeprecationNotice.php

nicolas-grekas · 2020-01-27T21:14:31Z

src/Symfony/Bridge/PhpUnit/DeprecationErrorHandler/DeprecationNotice.php

+        ++$this->count;
+    }
+
+    public function countsByCaller()


getCountsByCaller?

src/Symfony/Bridge/PhpUnit/Tests/DeprecationErrorHandler/DeprecationGroupTest.php

src/Symfony/Bridge/PhpUnit/Tests/DeprecationErrorHandler/DeprecationNoticeTest.php

nicolas-grekas · 2020-01-27T21:16:04Z

src/Symfony/Bridge/PhpUnit/Tests/DeprecationErrorHandler/lagging_vendor.phpt

@@ -33,7 +33,7 @@ require __DIR__.'/fake_vendor/acme/outdated-lib/outdated_file.php';

 ?>
 --EXPECTF--
-Remaining indirect deprecation notices (1)
+Indirect deprecation notices (1)


Why remove the "remaining" prefix? It made the output more friendly to me, I see no reasons to drop it.

It was my suggestion, check #35271 (comment). Remaining implies non-remaining ones are displayed somewhere else. Please help us see how it improves friendliness using that word.

When you're in the process of fixing deprecations - ie when you look at those reports, these are the "remaining deprecations" triggered by your code, until you fix them all. I don't read "remaining" as you and thus I don't see the confusion. Writing "Indirect deprecation notices (1)" feels too... direct. "Remaining" gives a direction to the cold fact, thus helps humans (because human don't know what to do with cold facts, but that's going off topic :) )

Why not „Deprecations you can fix“ and „Vendor Deprecations“?
Another wording is fine too, you get my point... hopefully 🙂

Unless there is an issue with the current wording, reducing merge conflicts is my priority, thanks for understanding :)

We can say the same about normal failing phpunit test cases. They are supposed/waiting to be fixed, but phpunit doesn't say "remainig failures". AFAIK nowhere else than phpunit bridge is this concept used like this.

Also, in practice user can't fix Indirect deprecations, but they are "remaining". So there is nothing to fix for user similar as "legacy" deprecations, but unlike legacy deprecations they are for some reason marked as "remaining".

To me whole remaining concept just makes things more confusing. But since we disagree, I guess this should be extracted to separate PR/issue.

@OskarStark also, "remaining" was also used for vendor deprecations :P

"vendor" deprecations are not a thing.

"indirect" are - YOU are always responsible for them, all of them. It could be because YOU decided to use this package or YOU didn't contribute the fix yet, or YOU are using the package in a deprecated way.

Let's stop the wrong narrative about "it's them, not me".

greg0ire · 2020-01-27T21:45:10Z

Can you please also update to the description of the PR so that it documents what this is about? I had to review in details to get it, the ppl that will write the doc and/or the blog post about this need help too.

Sorry for not doing that sooner, now that I wrote it guess the usage is a bit confusing, maybe the behavior should be changed, and it people should rather say which groups should be made quiet with verbose[indirect]=0?

nicolas-grekas · 2020-01-27T21:50:21Z

quiet=foo,bar,baz?

nicolas-grekas · 2020-01-27T21:50:55Z

or quiet[]=foo&quiet[]=bar, as you prefer :)

nicolas-grekas · 2020-01-27T21:53:00Z

ouch, I'm going again in the other alternative not suggested by @ostrolucky, sorry...
I feel like [key]=int might create confusion: "what does 2 mean?", that's why I proposed these.

greg0ire · 2020-01-27T21:55:55Z

Yeah, also, only one of 1 like right now or 0 (if we default to verbose groups) would ever be actually used. The syntax you proposed leaves less room for errors and confusion I think.

greg0ire · 2020-01-27T22:10:51Z

@nicolas-grekas what should happen if we have verbose=0&quiet[]=indirect? throw an exception telling the user not to use both options together?

nicolas-grekas · 2020-01-27T22:14:30Z

Nothing special to me.

greg0ire · 2020-01-27T22:49:58Z

@nicolas-grekas implemented :)

greg0ire · 2020-01-28T12:16:46Z

reducing merge conflicts is my priority,

@nicolas-grekas should I contribute the first commit to a lower branch first? It's a refactoring (and a big one).

nicolas-grekas · 2020-01-28T12:20:58Z

should I contribute the first commit to a lower branch first? It's a refactoring

nope, refacto are always for master: reducing the risk of introducing regressions has an even higher priority.

greg0ire · 2020-01-28T12:23:30Z

Ok, I just dropped the last commit (about "remaining")

nicolas-grekas

Some minor comments and good to me (I didn't run the PR so I'm counting on you to fix any issues we missed during review ;) )

nicolas-grekas · 2020-01-28T12:41:26Z

src/Symfony/Bridge/PhpUnit/DeprecationErrorHandler/Configuration.php

-     *                              whether to display a stack trace
-     * @param bool   $verboseOutput
+     * @param string $regex         Will be matched against messages,
+     *                              to decide whether to display a stack trace


unneeded diff

nicolas-grekas · 2020-01-28T12:43:42Z

src/Symfony/Bridge/PhpUnit/DeprecationErrorHandler/Configuration.php

@@ -72,7 +72,15 @@ private function __construct(array $thresholds = [], $regex = '', $verboseOutput
            }
        }
        $this->regex = $regex;
-        $this->verboseOutput = $verboseOutput;
+
+        $this->verboseOutput = array_fill_keys(['unsilenced', 'direct', 'indirect', 'self', 'other'], true);


Generally speaking, this still has runtime overhead, whereas declaring the resulting array is free. Also, I doubt this is more readable since it requires knowing what array_fill_keys() does.

nicolas-grekas · 2020-01-28T12:45:30Z

src/Symfony/Bridge/PhpUnit/DeprecationErrorHandler/DeprecationGroup.php

+
+namespace Symfony\Bridge\PhpUnit\DeprecationErrorHandler;
+
+final class DeprecationGroup


let's mark all new classes as @internal

nicolas-grekas · 2020-01-28T12:45:41Z

src/Symfony/Bridge/PhpUnit/DeprecationErrorHandler/DeprecationNotice.php

+final class DeprecationNotice
+{
+    /**
+     * @var int


greg0ire · 2020-01-28T18:15:11Z

Some minor comments and good to me (I didn't run the PR so I'm counting on you to fix any issues we missed during review ;) )

I just tested on SonataAdminBundle, and I couldn't find any issues 😛

fabpot · 2020-01-29T14:01:23Z

Thank you @greg0ire.

This PR was merged into the 5.1-dev branch. Discussion ---------- [Typo] Rename occurence to occurrence | Q | A | ------------- | --- | Branch? | master (5.1) | Bug fix? | yes (typo) | New feature? | no | Deprecations? | no | Tickets | Fix #36242 | License | MIT | Doc PR | Not needed This PR will fix correct spelling of methods containing the word occurrence implemented with PR : - [PHPUnitBridge] Improved deprecations display #35271 This changes are on master branch, because the original PR is a new feature for Symfony 5.1. Commits ------- 11f746a [Typo] Rename occurence to occurrence

carsonbot added Status: Needs Review PhpUnitBridge Feature labels Jan 8, 2020

greg0ire force-pushed the improved-deprecations-display branch 2 times, most recently from d803a11 to 5a5df3c Compare January 8, 2020 21:53

OskarStark reviewed Jan 9, 2020

View reviewed changes

nicolas-grekas added this to the next milestone Jan 9, 2020

ostrolucky reviewed Jan 16, 2020

View reviewed changes

greg0ire force-pushed the improved-deprecations-display branch 4 times, most recently from aafa219 to dbffb71 Compare January 20, 2020 18:50

greg0ire marked this pull request as ready for review January 20, 2020 19:02

ostrolucky suggested changes Jan 20, 2020

View reviewed changes

carsonbot added Status: Needs Work and removed Status: Needs Review labels Jan 20, 2020

greg0ire force-pushed the improved-deprecations-display branch 2 times, most recently from 9970c33 to 8d56650 Compare January 25, 2020 11:09

greg0ire requested a review from nicolas-grekas January 25, 2020 11:19

carsonbot added Status: Needs Review and removed Status: Needs Work labels Jan 25, 2020

greg0ire force-pushed the improved-deprecations-display branch from 8d56650 to e775a6d Compare January 26, 2020 11:50

nicolas-grekas reviewed Jan 27, 2020

View reviewed changes

greg0ire force-pushed the improved-deprecations-display branch from e775a6d to 5782a1e Compare January 27, 2020 22:08

greg0ire force-pushed the improved-deprecations-display branch 2 times, most recently from d4b9c32 to f56fc61 Compare January 27, 2020 22:41

greg0ire force-pushed the improved-deprecations-display branch from f56fc61 to dac6ad7 Compare January 28, 2020 12:23

nicolas-grekas approved these changes Jan 28, 2020

View reviewed changes

carsonbot added Status: Reviewed and removed Status: Needs Review labels Jan 28, 2020

greg0ire force-pushed the improved-deprecations-display branch from dac6ad7 to ef62255 Compare January 28, 2020 18:11

fabpot approved these changes Jan 29, 2020

View reviewed changes

[PHPUnitBridge] Improved deprecations display

c55a89e

fabpot force-pushed the improved-deprecations-display branch from ef62255 to c55a89e Compare January 29, 2020 14:00

fabpot mentioned this pull request Jan 29, 2020

[PHPUnitBridge] Improved deprecations display symfony/symfony-docs#13008

Closed

fabpot closed this in 7ecb5aa Jan 29, 2020

fabpot merged commit c55a89e into symfony:master Jan 29, 2020

This was referenced Mar 27, 2020

[Typo] Rename occurence to occurrence #36242

Closed

[Typo] Rename occurence to occurrence #36244

Merged

nicolas-grekas modified the milestones: next, 5.1 May 4, 2020

fabpot mentioned this pull request May 5, 2020

Release v5.1.0-BETA1 #36713

Merged

	public function countsByCaller()
	public function getMethodOccurences()


		namespace Symfony\Bridge\PhpUnit\DeprecationErrorHandler;

		final class DeprecationGroup

[PHPUnitBridge] Improved deprecations display #35271

[PHPUnitBridge] Improved deprecations display #35271

Conversation

greg0ire commented Jan 8, 2020 • edited Loading

OskarStark left a comment

Choose a reason for hiding this comment

ostrolucky left a comment • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

greg0ire commented Jan 20, 2020

ostrolucky commented Jan 20, 2020 • edited Loading

greg0ire commented Jan 26, 2020

nicolas-grekas left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

nicolas-grekas Jan 27, 2020 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

nicolas-grekas Jan 28, 2020 • edited Loading

Choose a reason for hiding this comment

greg0ire commented Jan 27, 2020

nicolas-grekas commented Jan 27, 2020

nicolas-grekas commented Jan 27, 2020

nicolas-grekas commented Jan 27, 2020 • edited Loading

greg0ire commented Jan 27, 2020

greg0ire commented Jan 27, 2020

nicolas-grekas commented Jan 27, 2020

greg0ire commented Jan 27, 2020

greg0ire commented Jan 28, 2020

nicolas-grekas commented Jan 28, 2020

greg0ire commented Jan 28, 2020

nicolas-grekas left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

greg0ire commented Jan 28, 2020

fabpot commented Jan 29, 2020

greg0ire commented Jan 8, 2020 •

edited

Loading

ostrolucky left a comment •

edited

Loading

ostrolucky commented Jan 20, 2020 •

edited

Loading

nicolas-grekas Jan 27, 2020 •

edited

Loading

nicolas-grekas Jan 28, 2020 •

edited

Loading

nicolas-grekas commented Jan 27, 2020 •

edited

Loading