Coming soon: second generation prototype of Hg forest consolidation
erik.joelsson at oracle.com
Tue Apr 11 10:00:37 UTC 2017
On 2017-04-11 07:26, David Holmes wrote:
> Hi Igor,
> I'm not Erik but this got my attention :)
> On 11/04/2017 4:55 AM, Igor Ignatyev wrote:
>> (moved to hotspot-dev alias)
>> Hi Erik,
>> I have a question about test/hotspot layout. when we started jep
>> 281, we deliberately selected not to use harness/framework names
>> in directory and the proposed the following layout for hotspot tests was
>>> For clarity the tests should not be mixed with existing jtreg tests;
>>> instead, they should be split in two directories. We propose to
>>> split the current jdk9/hotspot/test directory into two subdirectories:
>>> • jdk9/hotspot/test/java
>>> • jdk9/hotspot/test/native
>> now the consolidated repo has test/hotspot/jtreg and
>> test/hotspot/gtest and directories. hence I'm wondering why you have
>> renamed the directories, there were any problems w/ the originally
>> proposed layout?
> Mea culpa but I never paid much attention to JEP 281 so was unaware of
> the proposal for "java" and "native" subdirectories. If I had noticed
> it I would have objected as the names are not indicative of the
> contents. I think naming based on the testing framework makes much
> more sense: jtreg and gtest.
I was aware of the suggestion to use java and native, but I was never
part of the discussion that lead to that decision. I used that naming
for a long time in the prototype, but it didn't sit right with me
however since, as David points out, it does not reflect the contents
well. The jtreg tests consist of Java, C/C++ and shell script tests. The
gtest tests are C++. The term "native" is not clearly denoting one or
the other. Browsing this I wouldn't know if the native parts of jtreg
tests are supposed to be in that directory or not for example. There are
also a fair bit of framework specific parts of the implementations, like
jtreg tags, which make the tests very specific to their frameworks. I
just don't see the point in trying to hide that.
As I see it the problem with using the framework name would be if we one
day decided to change frameworks but didn't want to move the tests. I
would find this a very unlikely scenario. If we were to change
frameworks, it would be a lengthy process of converting test by test and
we would most likely want to move each converted test to a new directory
to clearly signify its conversion. So even in that case, a framework
based directory makes more sense.
> My 2c.
>> -- Igor
>>  JEP 281: HotSpot C++ Unit-Test Framework
>> http://openjdk.java.net/jeps/281 <http://openjdk.java.net/jeps/281>
>>> On Apr 10, 2017, at 9:42 AM, joe darcy <joe.darcy at oracle.com> wrote:
>>> Second generation prototype now available from
>>> Sources equivalent to JDK 9 b161.
>>> On 4/5/2017 9:44 AM, joe darcy wrote:
>>>> As previously discussed , we've been exploring consolidating the
>>>> many repositories in an OpenJDK Hg forest (jdk, langtools, hotspot,
>>>> ...) into a single repository for JDK 10. The initial prototype of
>>>> a consolidated repo , equivalent to JDK 9 build 138, was
>>>> published in October 2016.
>>>> Erik Joelsson has been working on a refined version of the
>>>> consolidation repo and we plan to publish this second generation
>>>> prototype in the near future. The new prototype presents a more
>>>> faithful view of the sources between tags for promoted builds and
>>>> provides a cleaner overall file layout.
>>>> The corresponding JEP  will be updated to reflect the details of
>>>> the new prototype in due course.
>>>>  "Looking ahead: proposed Hg forest consolidation for JDK 10,"
>>>>  http://hg.openjdk.java.net/jdk9/consol-proto/
>>>>  JEP 296: Consolidate the JDK Forest into a Single Repository
More information about the hotspot-dev