Current Status of O+ on Windows

11 years ago by Matt Ficken — view source

unread

I am testing O+ on Windows with php 5.3, 5.4, and 5.5.

I have uploaded my O+ test results here: http://windows.php.net/downloads/snaps/ostc/pftt/

Crashes have increased with 5.3, 5.4 and 5.5 TS builds with Apache. TS or NTS on CLI seem fine.

See http://windows.php.net/downloads/snaps/ostc/pftt/PHP_5_3/r61099f8/PHPT_CMP_PHP_5_3-r61099f8-TS-X86-VC9_Local-FileSystem_Apache-ModPHP_v_PHP_5_3-r61099f8-TS-X86-VC9_OptimizerPlus_Local-FileSystem_Apache-ModPHP.html and http://windows.php.net/downloads/snaps/ostc/pftt/PHP_5_4/r7c08232/PHPT_CMP_PHP_5_4-r7c08232-TS-X86-VC9_Local-FileSystem_Apache-ModPHP_v_PHP_5_4-r7c08232-TS-X86-VC9_OptimizerPlus_Local-FileSystem_Apache-ModPHP.html and http://windows.php.net/downloads/snaps/ostc/pftt/PHP_5_4/r7c08232/PHPT_CMP_PHP_5_4-r7c08232-TS-X86-VC9_Local-FileSystem_CLI_v_PHP_5_4-r7c08232-TS-X86-VC9_OptimizerPlus_Local-FileSystem_CLI.html

Crashes are up to 6 or 7 (in addition to the same 2 or 3) on 5.3 and 5.4. And with 5.5, crashes are up to 17 on CLI and 22 on Apache. 5.5 is compiled using VC11 whereas 5.4 and 5.3 are compiled using VC9 so that may be part of the issue with 5.5.

See http://windows.php.net/downloads/snaps/ostc/pftt/PHP_5_5/r0d65a85/PHPT_CMP_PHP_5_5-r0d65a85-TS-X86-VC11_Local-FileSystem_Apache-ModPHP_v_PHP_5_5-r0d65a85-TS-X86-VC11_OptimizerPlus_Local-FileSystem_Apache-ModPHP.html and http://windows.php.net/downloads/snaps/ostc/pftt/PHP_5_5/r0d65a85/PHPT_CMP_PHP_5_5-r0d65a85-TS-X86-VC11_Local-FileSystem_CLI_v_PHP_5_5-r0d65a85-TS-X86-VC11_OptimizerPlus_Local-FileSystem_CLI.html

Secondly, Reflection may be affected when running on Apache. PhpUnit uses reflection to run the actual test method. On Apache, all tests pass with O+ whereas on CLI or Apache without O+, I get ~38 errors and failures and some crashes (validated): Reflection seems broken only on Apache only with O+.

See http://windows.php.net/downloads/snaps/ostc/pftt/PHP_5_5/r0d65a85/PhpUnit_CMP_Symfony-Standard-2.1.8_PHP_5_5-r0d65a85-TS-X86-VC11_Local-FileSystem_Apache-ModPHP_v_PHP_5_5-r0d65a85-TS-X86-VC11_OptimizerPlus_Local-FileSystem_Apache-ModPHP.html and http://windows.php.net/downloads/snaps/ostc/pftt/PHP_5_5/r0d65a85/PhpUnit_CMP_Symfony-Standard-2.1.8_PHP_5_5-r0d65a85-TS-X86-VC11_Local-FileSystem_CLI_v_PHP_5_5-r0d65a85-TS-X86-VC11_OptimizerPlus_Local-FileSystem_CLI.html

PhpUnit uses ReflectionMethod::invokeArgs to actually invoke the test method to run the test, and if no exceptions are thrown, it assumes the test passes. I think at least some of the test methods aren't getting invoked (optimized out?) but no exception is thrown, so PhpUnit assumes the test passes.

I am still studying this Reflection issue. I will try to create a PHPT test for this reflection+O+ issue.

Regards
-M

11 years ago by Christopher Jones — view source

unread

I am testing O+ on Windows with php 5.3, 5.4, and 5.5.

I have uploaded my O+ test results here: http://windows.php.net/downloads/snaps/ostc/pftt/

Can you put your O+ php.ini settings up there as well?
Did you experiment with any options?
Putting the date of your O+ snapshot would also be handy.

Chris

--
christopher.jones@oracle.com http://twitter.com/ghrd
Newly updated, free PHP & Oracle book:
http://www.oracle.com/technetwork/topics/php/underground-php-oracle-manual-098250.html

11 years ago by Pierre Joye — view source

unread

On Sat, Mar 2, 2013 at 1:30 AM, Christopher Jones
christopher.jones@oracle.com wrote:

Can you put your O+ php.ini settings up there as well?

All default, especially for symfony tests, which must have the comment
setting enabled (while some tests still fail even if the settings are
enabled).

Did you experiment with any options?
Putting the date of your O+ snapshot would also be handy.

Latest from here are used:
http://windows.php.net/downloads/pecl/snaps/Optimizer/7.0.0-dev/

Dates are included. It would be nice to have it in the report as well,
but we always use latest revision. It would however be much easier if
there were PECL releases.

Cheers,

Pierre

@pierrejoye

11 years ago by Ferenc Kovacs — view source

unread

Did you experiment with any options?
Putting the date of your O+ snapshot would also be handy.

Latest from here are used:
http://windows.php.net/downloads/pecl/snaps/Optimizer/7.0.0-dev/

Dates are included. It would be nice to have it in the report as well,
but we always use latest revision. It would however be much easier if
there were PECL releases.

+1
Dmitry, is there any objection against creating a pecl release?
Could you tag the first version?

--
Ferenc Kovács
@Tyr43l - http://tyrael.hu

11 years ago by Zeev Suraski — view source

unread

-----Original Message-----
From: Ferenc Kovacs [mailto:tyra3l@gmail.com]
Sent: Saturday, March 02, 2013 10:15 AM
To: Pierre Joye; Dmitry Stogov
Cc: Christopher Jones; Matt Ficken; internals@lists.php.net
Subject: Re: [PHP-DEV] Current Status of O+ on Windows

Did you experiment with any options?
Putting the date of your O+ snapshot would also be handy.

Latest from here are used:
http://windows.php.net/downloads/pecl/snaps/Optimizer/7.0.0-dev/

Dates are included. It would be nice to have it in the report as well,
but we always use latest revision. It would however be much easier if
there were PECL releases.

+1
Dmitry, is there any objection against creating a pecl release?
Could you tag the first version?

The current vote that's going on right now deals with putting the extension
into PHP itself. If that happens (which seems awfully likely at this
point), why do we need it in PECL? We'd very much rather invest our very
limited cycles into the code itself. We're probably roughly a week away
from having builds as a part of the PHP 5.5 snaps.

Zeev

11 years ago by Ferenc Kovacs — view source

unread

-----Original Message-----
From: Ferenc Kovacs [mailto:tyra3l@gmail.com]
Sent: Saturday, March 02, 2013 10:15 AM
To: Pierre Joye; Dmitry Stogov
Cc: Christopher Jones; Matt Ficken; internals@lists.php.net
Subject: Re: [PHP-DEV] Current Status of O+ on Windows

Did you experiment with any options?
Putting the date of your O+ snapshot would also be handy.

Latest from here are used:
http://windows.php.net/downloads/pecl/snaps/Optimizer/7.0.0-dev/

Dates are included. It would be nice to have it in the report as well,
but we always use latest revision. It would however be much easier if
there were PECL releases.

+1
Dmitry, is there any objection against creating a pecl release?
Could you tag the first version?

The current vote that's going on right now deals with putting the extension
into PHP itself. If that happens (which seems awfully likely at this
point), why do we need it in PECL? We'd very much rather invest our very
limited cycles into the code itself. We're probably roughly a week away
from having builds as a part of the PHP 5.5 snaps.

Zeev

I see.
so no O+ for <5.5?
having a pecl release would be a small amount of work, which I would glad
to help with.

--
Ferenc Kovács
@Tyr43l - http://tyrael.hu

11 years ago by Reeze Xia — view source

unread

+1 s, Switch to a new opcode cacher is much easier than update PHP,
and ZO+ is already compatible with <5.5. we could use ZO+ as soon as possible.

reeze | reeze.cn

已使用 Sparrow (http://www.sparrowmailapp.com/?sig)

在 2013年3月2日星期六，下午4:43，Ferenc Kovacs 写道：

-----Original Message-----
From: Ferenc Kovacs [mailto:tyra3l@gmail.com]
Sent: Saturday, March 02, 2013 10:15 AM
To: Pierre Joye; Dmitry Stogov
Cc: Christopher Jones; Matt Ficken; internals@lists.php.net (mailto:internals@lists.php.net)
Subject: Re: [PHP-DEV] Current Status of O+ on Windows

Did you experiment with any options?
Putting the date of your O+ snapshot would also be handy.

Latest from here are used:
http://windows.php.net/downloads/pecl/snaps/Optimizer/7.0.0-dev/

Dates are included. It would be nice to have it in the report as well,
but we always use latest revision. It would however be much easier if
there were PECL releases.

+1
Dmitry, is there any objection against creating a pecl release?
Could you tag the first version?

The current vote that's going on right now deals with putting the extension
into PHP itself. If that happens (which seems awfully likely at this
point), why do we need it in PECL? We'd very much rather invest our very
limited cycles into the code itself. We're probably roughly a week away
from having builds as a part of the PHP 5.5 snaps.

Zeev

I see.
so no O+ for <5.5?
having a pecl release would be a small amount of work, which I would glad
to help with.

--
Ferenc Kovács
@Tyr43l - http://tyrael.hu

11 years ago by Terry Ellison — view source

unread

The current vote that's going on right now deals with putting the extension
into PHP itself. If that happens (which seems awfully likely at this
point), why do we need it in PECL?
My response to your Q is that there is probably going to be quite a lot
of interest in an O+ package that is usable with PHP 5.3 and 5.4.
Surely a PECL package will have a quicker uptake terms of getting it out
into the wider PHP developers community and into production, especially
if the main Linux distros add a precompiled php5-optimizer-plus package
(or whatever their naming convention is).

Would you see such O+ support for the existing supported versions best
done through the PECL route or swept up into a maintenance dot release?

Regards Terry

11 years ago by Matt Ficken — view source

unread

Earlier I had posted about testing O+ with Apache using Symfony's PhpUnit tests...
I had an issue where it seems that test methods (which PhpUnit accesses using ReflectionMethod) were getting optimized out and replaced with an empty method (so all PhpUnit tests passed).
I have now written a PHPT to try to reproduce this issue (attached in case it helps), which passes/doesn’t have this
issue.

I don’t think this issue is full
resolved/closed, but it’s a low-priority edge case that can be fixed later.

Symfony runs with O+(see O+
README for proper config)…. This issue doesn’t affect a normal Symfony app
running on Apache with O+... users aren't likely to have this problem.
If users do have this problem, I have found that turning off all optimization passes except pass 3 and pass 10 (using the optimization_level directive) seems to fix this problem.

The reason I found this problem is that I
run multiple PhpUnit tests on each Apache instance, one test is run at a time,
but multiple tests are run against each instance/process during its lifetime.
This is a little more accurate (apache doesn’t reboot between each HTTP
request) and makes testing much faster, otherwise I have to start and stop
apache between each test, which is really slow.

I have confirmed this by
restarting apache between each PhpUnit test, and I get about
the same errors and failures as I do without O+. Additionally, I get a bunch of
crashes, which I will be investigating further.

Regardless of the crashes, this
issue is not critical/low priority now since all the PhpUnit tests actually now
run (reflectionmethods aren’t getting optimized out).

Regards

-Matt

Date: Sat, 2 Mar 2013 09:43:34 +0100
From: tyra3l@gmail.com
To: zeev@zend.com
CC: internals@lists.php.net; dmitry@zend.com
Subject: Re: [PHP-DEV] Current Status of O+ on Windows

-----Original Message-----
From: Ferenc Kovacs [mailto:tyra3l@gmail.com]
Sent: Saturday, March 02, 2013 10:15 AM
To: Pierre Joye; Dmitry Stogov
Cc: Christopher Jones; Matt Ficken; internals@lists.php.net
Subject: Re: [PHP-DEV] Current Status of O+ on Windows

Did you experiment with any options?
Putting the date of your O+ snapshot would also be handy.

Latest from here are used:
http://windows.php.net/downloads/pecl/snaps/Optimizer/7.0.0-dev/

Dates are included. It would be nice to have it in the report as well,
but we always use latest revision. It would however be much easier if
there were PECL releases.

+1
Dmitry, is there any objection against creating a pecl release?
Could you tag the first version?

The current vote that's going on right now deals with putting the extension
into PHP itself. If that happens (which seems awfully likely at this
point), why do we need it in PECL? We'd very much rather invest our very
limited cycles into the code itself. We're probably roughly a week away
from having builds as a part of the PHP 5.5 snaps.

Zeev

I see.
so no O+ for <5.5?
having a pecl release would be a small amount of work, which I would glad
to help with.

--
Ferenc Kovács
@Tyr43l - http://tyrael.hu

Current Status of O+ on Windows

Cheers,

+1 s, Switch to a new opcode cacher is much easier than update PHP, and ZO+ is already compatible with <5.5. we could use ZO+ as soon as possible.

+1 s, Switch to a new opcode cacher is much easier than update PHP,
and ZO+ is already compatible with <5.5. we could use ZO+ as soon as possible.