Opened 4 years ago
Last modified 3 months ago
#51383 reviewing defect (bug)
PHP error body class added for suppressed errors
Reported by: | desrosj | Owned by: | |
---|---|---|---|
Milestone: | Future Release | Priority: | normal |
Severity: | normal | Version: | 5.5 |
Component: | Administration | Keywords: | has-patch has-unit-tests |
Focuses: | Cc: |
Description
When a PHP error occurs in the admin that is suppressed using the @
operator, the .php-error
class is still added to the <body>
even though the error will not be output.
To reproduce, use error suppression on code that invokes an E_WARNING
level error. I discovered this with code calling the mkdir()
method within the WP_Filesystem_Direct
class for a pre-existing directory.
@mkdir( ABSPATH . 'wp-content/themes/' );
should reproduce the issue after ensuring nothing else is actively capturing and processing PHP errors (such as Query Monitor).
Attachments (4)
Change History (43)
#2
@
4 years ago
- Keywords has-patch added; needs-patch removed
Hello there, I have added the E_WARNING to the condition, and when the error is silenced, the empty space is not present however, if the error is not silenced part of the message appears under the menu. I will tweak it a bit more, but if anyone has another idea, I am open to suggestions. I have tried using,
! in_array( $error_get_last['type'], array( E_NOTICE, E_WARNING )
, but the second condition causes the empty space to be present if the error is silenced elsewhere.
This ticket was mentioned in Slack in #core by audrasjb. View the logs.
4 years ago
#5
@
4 years ago
During today’s bug scrub, @SergeyBiryukov shared some concerns about the proposed approach. By the way, I tested the patch and it looks like it fixes the issue.
#6
@
4 years ago
It seems patch not consider last space in the code
&& ( E_NOTICE !== $error_get_last['type'] || 'wp-config.php' !== wp_basename( $error_get_last['file'] ) )
This ticket was mentioned in Slack in #core by desrosj. View the logs.
4 years ago
#8
@
4 years ago
- Keywords commit added
@SergeyBiryukov, any other thoughts on this one? Looks good to me.
#9
@
4 years ago
- Milestone changed from 5.5.2 to 5.6
Would like some more time to review this, moving to 5.6 for now.
#11
@
4 years ago
I just encountered the issue in the real world.
I applied the patch and it worked fine on my side. @SergeyBiryukov not marking this for commit
as you wanted to take the time to deeply review the patch, but it looks good to go on my side :)
#13
@
4 years ago
Would love to get this in 5.7, but it would need to be merged inside a week. Think we can make that happen @SergeyBiryukov ?
This ticket was mentioned in Slack in #core by hellofromtonya. View the logs.
4 years ago
#15
@
4 years ago
- Milestone changed from 5.7 to 5.8
5.7 RC1 is tomorrow. Per scrub discussion today with @SergeyBiryukov:
I think the patch might fix the issue at hand, but I'm not sure that code still does what it's intended to do then :slightly_smiling_face: Since it's not specific to 5.7, I think it can be punted to 5.8.
Moving this ticket to 5.8.
This ticket was mentioned in Slack in #core by chaion07. View the logs.
4 years ago
#17
@
3 years ago
- Milestone changed from 5.8 to 5.9
I agree with Sergey that a deeper review is needed to ensure the original intent remains. Also should revisit his suggestions here:
In a future release, we could try something like:
- Saving the value of
error_reporting()
before it's affected bywp_debug_mode()
.. - Using that to only print the class if we determine that the error was displayed according to that setting.
As the issue is not due to 5.7 or 5.8 and today is 5.8 Beta 1, punting to 5.9.
Adding to my tasks list to help out for resolution landing in 5.9.
This ticket was mentioned in Slack in #core by audrasjb. View the logs.
3 years ago
#19
@
3 years ago
- Owner changed from SergeyBiryukov to hellofromTonya
Reassigning to me to ensure it gets into my review queue.
#20
@
3 years ago
FWIW, I just checked the last proposed patch and it still applies cleanly against trunk @hellofromTonya :)
This ticket was mentioned in Slack in #core by audrasjb. View the logs.
3 years ago
#22
@
3 years ago
- Keywords needs-testing added
- Milestone changed from 5.9 to 6.0
Likely won't get time to review with 5.9 Beta 1 in 5 days. Though this fix is straightforward, there's history that needs to be considered to ensure the fix retains the expected behavior and fixes the reported issue. It'll testing.
This ticket was mentioned in Slack in #core by costdev. View the logs.
3 years ago
#24
@
3 years ago
Per the discussion in the bug scrub, pinging @hellofromTonya to see if you've had time/have time to do the review on this?
#25
@
3 years ago
- Keywords needs-refresh added
The patch doesn't apply anymore against trunk.
I'm adding a PR right now :)
This ticket was mentioned in PR #2490 on WordPress/wordpress-develop by audrasjb.
3 years ago
#26
- Keywords needs-refresh removed
Trac ticket: https://core.trac.wordpress.org/ticket/51383
This ticket was mentioned in Slack in #core by chaion07. View the logs.
3 years ago
#28
@
3 years ago
@SergeyBiryukov @hellofromTonya The change works on my side, but it looks like you both wanted to have a second pair on eyes on this. Are you available to double check this PR?
#29
@
3 years ago
- Keywords needs-refresh added; needs-testing removed
I don't think the patch is correct.
The PHP class ought to be added if the warning is unsuppressed but with the linked pull request the class is never displayed.
As a proof of concept, apply the patch and add trigger_error( 'some warning', E_WARNING );
prior to the definition of $error_get_last
.
The warning will be displayed but the class won't be added to the body
element.
Developer types: remember to deactivate Query Monitor while testing this, it sets an error handler that may trigger a few minutes confusion.
#30
@
3 years ago
- Milestone changed from 6.0 to Future Release
- Owner hellofromTonya deleted
With 6.0 RC1 tomorrow, it's too late for this ticket in the 6.0 cycle. As the ticket has been kicked down the road since 5.5.2 and there's still work pending to figure out a solution, moving this to Future Release
. Also removing myself as owner.
#32
@
2 years ago
Committers: please give props to @hztyfoon
They added some testing notes to the duplicate ticket, #56947.
#34
follow-up:
↓ 35
@
23 months ago
Example related to 51383.2.diff:
<?php // Ignore suppressed deprecation notices. add_filter( 'pre_admin_php_error_class', function ( $print, $error ) { return ( $print && E_DEPRECATED !== $error['type'] ) }, 10, 2 );
#35
in reply to:
↑ 34
@
18 months ago
@mcaskill The last provided solution is working fine and in the newer versions of WordPress the same class indicates to display the PHP errors in a proper way.
This ticket was mentioned in Slack in #core-test by oglekler. View the logs.
15 months ago
This ticket was mentioned in Slack in #core by oglekler. View the logs.
4 months ago
This ticket was mentioned in PR #7256 on WordPress/wordpress-develop by @akirk.
3 months ago
#39
- Keywords has-unit-tests added; needs-refresh removed
In PHP 8.1, the admin_header will gets the php-error
CSS class added for the suppressed notice for @ini_set( 'auto_detect_line_endings', 1 );
This adds code so that we can do without the ini setting and still handle old-style \r
line-terminated files. This also adds a unit test for this.
Trac ticket: https://core.trac.wordpress.org/ticket/51383
Placing this in the 5.5.2 milestone as this would be a good to fix to backport.
Also, I forgot to include why this is undesirable. The
.php-error
class ensures PHP errors are not hidden behind the admin bar. But since the error is suppressed, it just results in a blank space at the top of the page.