From martinrb at google.com  Thu Feb  7 17:28:21 2019
From: martinrb at google.com (Martin Buchholz)
Date: Thu, 7 Feb 2019 09:28:21 -0800
Subject: [8u] RFR (XS) 8218613: [TESTBUG] runtime/ErrorHandling tests are
 building incorrect testlibrary classes
In-Reply-To: <cb15dd09-bdc4-c0eb-8b7f-17a7adff8c19@oracle.com>
References: <12824be8-7d91-f832-51e5-491c7b06b9a8@redhat.com>
 <cb15dd09-bdc4-c0eb-8b7f-17a7adff8c19@oracle.com>
Message-ID: <CA+kOe09kHzi8u_m13qwqKabcwDBNy+beJmgffWjWa1pxoXaqew@mail.gmail.com>

On Thu, Feb 7, 2019 at 4:33 AM David Holmes <david.holmes at oracle.com> wrote:

> On 7/02/2019 9:52 pm, Aleksey Shipilev wrote:
> > Bug:
> >    https://bugs.openjdk.java.net/browse/JDK-8218613
> >
> > This is botched backport to 8u, which fails the jtreg tests now in 8u.
> Note the fix is not the
> > backport, but the actual 8u-specific fix.
>
> IIRC the incorrect @build is not a fatal error for earlier jtreg
> version. AFAICS we still use jtreg 4.1 for testing 8u internally.
>

This surprised me.  Of course, the incentive to update to new jtreg
versions is generally because jdk-latest needs it, but we've always
successfully adopted any newer jtreg to test older jdks as well.
(but we don't run as many tests as you do)
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://mail.openjdk.java.net/pipermail/jtreg-use/attachments/20190207/80c8fa6c/attachment.html>

From arno.zeller at sap.com  Fri Feb  8 16:03:51 2019
From: arno.zeller at sap.com (Zeller, Arno)
Date: Fri, 8 Feb 2019 16:03:51 +0000
Subject: Ensure a test runs without other tests in parallel
Message-ID: <664ff22ffeff4606a4662155e577e2f9@sap.com>

Hello,



while I am investigating a problem with a jtreg test that fails because the test machine is running out of native memory I hoped to find a tag that ensures the test will run with no other test running in parallel - even when I start the jtreg harness with -conc:<x>  and  x > 1.

There are a few test that start the VM with a big heap (4 or 8GB) heap and if these might run concurrent it will always lead to problems. There might be other test that use special resources that might be limited.



I looked at the tag specification (http://openjdk.java.net/jtreg/tag-spec.html) but did not find a tag for this.

What I found is the following keyword for TEST.properties files:

exclusiveAccess.dirs <directories>
Directories containing tests which should not be run concurrently at the same time as other tests. A harness that supports concurrent test execution should ensure that all tests in the specified directories are run when no other tests in the same directory is being run.



But as far as I understand this only ensures that test in the same directory are not run concurrently - still other test from a different directory could run concurrently and might compete for limited resources. Did I get this right?



Is there another possibility to specify that test should run alone?



Best regards,

Arno

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://mail.openjdk.java.net/pipermail/jtreg-use/attachments/20190208/00887f6e/attachment.html>

From aph at redhat.com  Fri Feb  8 13:52:14 2019
From: aph at redhat.com (Andrew Haley)
Date: Fri, 8 Feb 2019 13:52:14 +0000
Subject: [8u] RFR (XS) 8218613: [TESTBUG] runtime/ErrorHandling tests are
 building incorrect testlibrary classes
In-Reply-To: <CA+kOe09kHzi8u_m13qwqKabcwDBNy+beJmgffWjWa1pxoXaqew@mail.gmail.com>
References: <12824be8-7d91-f832-51e5-491c7b06b9a8@redhat.com>
 <cb15dd09-bdc4-c0eb-8b7f-17a7adff8c19@oracle.com>
 <CA+kOe09kHzi8u_m13qwqKabcwDBNy+beJmgffWjWa1pxoXaqew@mail.gmail.com>
Message-ID: <224a577d-cd12-73e7-75c6-069972b97211@redhat.com>

On 2/7/19 5:28 PM, Martin Buchholz wrote:
> On Thu, Feb 7, 2019 at 4:33 AM David Holmes <david.holmes at oracle.com> wrote:
> 
>> On 7/02/2019 9:52 pm, Aleksey Shipilev wrote:
>>> Bug:
>>>    https://bugs.openjdk.java.net/browse/JDK-8218613
>>>
>>> This is botched backport to 8u, which fails the jtreg tests now in 8u.
>> Note the fix is not the
>>> backport, but the actual 8u-specific fix.
>>
>> IIRC the incorrect @build is not a fatal error for earlier jtreg
>> version. AFAICS we still use jtreg 4.1 for testing 8u internally.
>>
> 
> This surprised me.  Of course, the incentive to update to new jtreg
> versions is generally because jdk-latest needs it, but we've always
> successfully adopted any newer jtreg to test older jdks as well.
> (but we don't run as many tests as you do)

I want to use the current jtreg for everything if at all possible.

-- 
Andrew Haley
Java Platform Lead Engineer
Red Hat UK Ltd. <https://www.redhat.com>
EAC8 43EB D3EF DB98 CC77 2FAD A5CD 6035 332F A671

From jonathan.gibbons at oracle.com  Fri Feb  8 16:09:27 2019
From: jonathan.gibbons at oracle.com (Jonathan Gibbons)
Date: Fri, 8 Feb 2019 08:09:27 -0800
Subject: Ensure a test runs without other tests in parallel
In-Reply-To: <664ff22ffeff4606a4662155e577e2f9@sap.com>
References: <664ff22ffeff4606a4662155e577e2f9@sap.com>
Message-ID: <96f8a6f5-604a-6aa8-bb02-7470a1089b12@oracle.com>

Arno,

exclusiveAccess.dirs is what you are looking for. The text in the 
description should be improved to indicate that the lock applies to all 
tests in all the nominated directories.

For example, this would be a reasonable update to the spec:

- should ensure that all tests in the specified directories are run when 
no other tests in the same directory is being run.
+ should ensure that all tests in the specified directories are run when 
no other tests in any of the given directories are being run.

-- Jon

On 2/8/19 8:03 AM, Zeller, Arno wrote:
>
> Hello,
>
> while I am investigating a problem with a jtreg test that fails 
> because the test machine is running out of native memory I hoped to 
> find a tag that ensures the test will run with no other test running 
> in parallel ? even when I start the jtreg harness with -conc:<x> ?and? 
> x > 1.
>
> There are a few test that start the VM with a big heap (4 or 8GB) heap 
> and if these might run concurrent it will always lead to problems. 
> There might be other test that use special resources that might be 
> limited.
>
> I looked at the tag specification 
> (http://openjdk.java.net/jtreg/tag-spec.html) but did not find a tag 
> for this.
>
> What I found is the following keyword for TEST.properties files:
>
> exclusiveAccess.dirs /<directories>/
>
> Directories containing tests which should not be run concurrently at 
> the same time as other tests. A harness that supports concurrent test 
> execution should ensure that all tests in the specified directories 
> are run when no other tests in the same directory is being run.
>
> But as far as I understand this only ensures that test in the same 
> directory are not run concurrently - still other test from a different 
> directory could run concurrently and might compete for limited 
> resources. Did I get this right?
>
> Is there another possibility to specify that test should run alone?
>
> Best regards,
>
> Arno
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://mail.openjdk.java.net/pipermail/jtreg-use/attachments/20190208/a9191d55/attachment.html>

From aph at redhat.com  Tue Feb 12 14:21:52 2019
From: aph at redhat.com (Andrew Haley)
Date: Tue, 12 Feb 2019 14:21:52 +0000
Subject: [8u] RFR (XS) 8218613: [TESTBUG] runtime/ErrorHandling tests are
 building incorrect testlibrary classes
In-Reply-To: <ead53ffb-3405-47c9-8e5d-2f697f63523c@redhat.com>
References: <12824be8-7d91-f832-51e5-491c7b06b9a8@redhat.com>
 <cb15dd09-bdc4-c0eb-8b7f-17a7adff8c19@oracle.com>
 <CA+kOe09kHzi8u_m13qwqKabcwDBNy+beJmgffWjWa1pxoXaqew@mail.gmail.com>
 <224a577d-cd12-73e7-75c6-069972b97211@redhat.com>
 <ead53ffb-3405-47c9-8e5d-2f697f63523c@redhat.com>
Message-ID: <d307bcb3-072d-3492-2743-939208526612@redhat.com>

On 2/12/19 1:43 PM, Aleksey Shipilev wrote:
> On 2/8/19 2:52 PM, Andrew Haley wrote:
>> On 2/7/19 5:28 PM, Martin Buchholz wrote:
>>> On Thu, Feb 7, 2019 at 4:33 AM David Holmes <david.holmes at oracle.com> wrote:
>>>
>>>> On 7/02/2019 9:52 pm, Aleksey Shipilev wrote:
>>>>> Bug:
>>>>>    https://bugs.openjdk.java.net/browse/JDK-8218613
>>>>>
>>>>> This is botched backport to 8u, which fails the jtreg tests now in 8u.
>>>> Note the fix is not the
>>>>> backport, but the actual 8u-specific fix.
>>>>
>>>> IIRC the incorrect @build is not a fatal error for earlier jtreg
>>>> version. AFAICS we still use jtreg 4.1 for testing 8u internally.
>>>>
>>>
>>> This surprised me.  Of course, the incentive to update to new jtreg
>>> versions is generally because jdk-latest needs it, but we've always
>>> successfully adopted any newer jtreg to test older jdks as well.
>>> (but we don't run as many tests as you do)
>>
>> I want to use the current jtreg for everything if at all possible.
> 
> That is my goal as well. Can I have the push approval, please?

OK.

-- 
Andrew Haley
Java Platform Lead Engineer
Red Hat UK Ltd. <https://www.redhat.com>
EAC8 43EB D3EF DB98 CC77 2FAD A5CD 6035 332F A671

From jonathan.gibbons at oracle.com  Wed Feb 13 00:38:37 2019
From: jonathan.gibbons at oracle.com (Jonathan Gibbons)
Date: Tue, 12 Feb 2019 16:38:37 -0800
Subject: [8u] RFR (XS) 8218613: [TESTBUG] runtime/ErrorHandling tests are
 building incorrect testlibrary classes
In-Reply-To: <224a577d-cd12-73e7-75c6-069972b97211@redhat.com>
References: <12824be8-7d91-f832-51e5-491c7b06b9a8@redhat.com>
 <cb15dd09-bdc4-c0eb-8b7f-17a7adff8c19@oracle.com>
 <CA+kOe09kHzi8u_m13qwqKabcwDBNy+beJmgffWjWa1pxoXaqew@mail.gmail.com>
 <224a577d-cd12-73e7-75c6-069972b97211@redhat.com>
Message-ID: <629ca322-dcd7-52ed-6840-32c3cad27151@oracle.com>


On 2/8/19 5:52 AM, Andrew Haley wrote:
> On 2/7/19 5:28 PM, Martin Buchholz wrote:
>> On Thu, Feb 7, 2019 at 4:33 AM David Holmes <david.holmes at oracle.com> wrote:
>>
>>> On 7/02/2019 9:52 pm, Aleksey Shipilev wrote:
>>>> Bug:
>>>>     https://bugs.openjdk.java.net/browse/JDK-8218613
>>>>
>>>> This is botched backport to 8u, which fails the jtreg tests now in 8u.
>>> Note the fix is not the
>>>> backport, but the actual 8u-specific fix.
>>> IIRC the incorrect @build is not a fatal error for earlier jtreg
>>> version. AFAICS we still use jtreg 4.1 for testing 8u internally.
>>>
>> This surprised me.  Of course, the incentive to update to new jtreg
>> versions is generally because jdk-latest needs it, but we've always
>> successfully adopted any newer jtreg to test older jdks as well.
>> (but we don't run as many tests as you do)
> I want to use the current jtreg for everything if at all possible.
>

Folk,

It is a general goal for jtreg to support use on older systems, so I 
have a question regarding that aspect ... what is the oldest JDK for 
which you want to use jtreg (jdk7u?, jdk8u etc) and how do you run jtreg 
... using the Makefiles, or in some custom setup? There's a hidden 
aspect to that last question, which is, what JDK do you use to run jtreg 
when testing these older platforms?

-- Jon


From shade at redhat.com  Tue Feb 12 13:43:01 2019
From: shade at redhat.com (Aleksey Shipilev)
Date: Tue, 12 Feb 2019 14:43:01 +0100
Subject: [8u] RFR (XS) 8218613: [TESTBUG] runtime/ErrorHandling tests are
 building incorrect testlibrary classes
In-Reply-To: <224a577d-cd12-73e7-75c6-069972b97211@redhat.com>
References: <12824be8-7d91-f832-51e5-491c7b06b9a8@redhat.com>
 <cb15dd09-bdc4-c0eb-8b7f-17a7adff8c19@oracle.com>
 <CA+kOe09kHzi8u_m13qwqKabcwDBNy+beJmgffWjWa1pxoXaqew@mail.gmail.com>
 <224a577d-cd12-73e7-75c6-069972b97211@redhat.com>
Message-ID: <ead53ffb-3405-47c9-8e5d-2f697f63523c@redhat.com>

On 2/8/19 2:52 PM, Andrew Haley wrote:
> On 2/7/19 5:28 PM, Martin Buchholz wrote:
>> On Thu, Feb 7, 2019 at 4:33 AM David Holmes <david.holmes at oracle.com> wrote:
>>
>>> On 7/02/2019 9:52 pm, Aleksey Shipilev wrote:
>>>> Bug:
>>>>    https://bugs.openjdk.java.net/browse/JDK-8218613
>>>>
>>>> This is botched backport to 8u, which fails the jtreg tests now in 8u.
>>> Note the fix is not the
>>>> backport, but the actual 8u-specific fix.
>>>
>>> IIRC the incorrect @build is not a fatal error for earlier jtreg
>>> version. AFAICS we still use jtreg 4.1 for testing 8u internally.
>>>
>>
>> This surprised me.  Of course, the incentive to update to new jtreg
>> versions is generally because jdk-latest needs it, but we've always
>> successfully adopted any newer jtreg to test older jdks as well.
>> (but we don't run as many tests as you do)
> 
> I want to use the current jtreg for everything if at all possible.

That is my goal as well. Can I have the push approval, please?

-Aleksey


-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 833 bytes
Desc: OpenPGP digital signature
URL: <https://mail.openjdk.java.net/pipermail/jtreg-use/attachments/20190212/c89eab42/signature.asc>

From shade at redhat.com  Wed Feb 13 09:10:12 2019
From: shade at redhat.com (Aleksey Shipilev)
Date: Wed, 13 Feb 2019 10:10:12 +0100
Subject: [8u] RFR (XS) 8218613: [TESTBUG] runtime/ErrorHandling tests are
 building incorrect testlibrary classes
In-Reply-To: <629ca322-dcd7-52ed-6840-32c3cad27151@oracle.com>
References: <12824be8-7d91-f832-51e5-491c7b06b9a8@redhat.com>
 <cb15dd09-bdc4-c0eb-8b7f-17a7adff8c19@oracle.com>
 <CA+kOe09kHzi8u_m13qwqKabcwDBNy+beJmgffWjWa1pxoXaqew@mail.gmail.com>
 <224a577d-cd12-73e7-75c6-069972b97211@redhat.com>
 <629ca322-dcd7-52ed-6840-32c3cad27151@oracle.com>
Message-ID: <79b48687-691c-81cf-c2ac-74b175bbf730@redhat.com>

On 2/13/19 1:38 AM, Jonathan Gibbons wrote:
> It is a general goal for jtreg to support use on older systems, so I have a question regarding that
> aspect ... what is the oldest JDK for which you want to use jtreg (jdk7u?, jdk8u etc) and how do you
> run jtreg ... using the Makefiles, or in some custom setup? There's a hidden aspect to that last
> question, which is, what JDK do you use to run jtreg when testing these older platforms?

I run "make test", "make run-test" with whatever boot jdk is accepted in jdk9+. jdk8u does not have
lots of test definitions in TEST.groups (we would rectify that hopefully soon), so executing the
jtreg as standalone binary with system Java (8u), passing the $repo/test directory to it.

I think the minimal supported version should be jdk7u.

-Aleksey


-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 833 bytes
Desc: OpenPGP digital signature
URL: <https://mail.openjdk.java.net/pipermail/jtreg-use/attachments/20190213/a4c0737c/signature.asc>

From jonathan.gibbons at oracle.com  Thu Feb 21 17:08:10 2019
From: jonathan.gibbons at oracle.com (Jonathan Gibbons)
Date: Thu, 21 Feb 2019 09:08:10 -0800
Subject: jtreg 4.2-b14
Message-ID: <12528e93-df6e-6bb9-3d45-82c067b18c0b@oracle.com>

Some of you may have noticed that a new version of jtreg is available 
(jtreg4.2-b14) and is now being required in various JDK test suites.

There are two note-worthy features in this update.

 1. Action tags (like `@compile` or `@run main`) now support variable
    substitution in command arguments, using the syntax `${name}`. The
    set of names is the same as that for the `@requires` tag.? The
    feature is enabled by default for test suites that specify the use
    of jtreg4.2-b14 or later.

 2. Support is now available for using Windows Subsystem for Linux (WSL)
    as an alternative to Cygwin to run shell tests on Windows 10. This
    complements the recent new ability to build OpenJDK using WSL. Note
    that shell tests will also need to be updated in order to use this
    feature: a discussion of the issues and some recommendations have
    been posted on the OpenJDK website:
    https://openjdk.java.net/jtreg/shellTests.html

Many of the updates have been changes to the documentation, such as the 
jtreg FAQ:
https://openjdk.java.net/jtreg/faq.html

A full list of changes can be found here:
http://hg.openjdk.java.net/code-tools/jtreg/search/?rev=%22jtreg4.2-b13%22%3A%22jtreg4.2-b14%22&revcount=80

-- Jon

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://mail.openjdk.java.net/pipermail/jtreg-use/attachments/20190221/a3e2d88b/attachment.html>