Request for assistance: Verify and update mailing list filter rules for jdk/jdk in preparation for Skara transition
david.holmes at oracle.com
Thu Aug 27 13:21:13 UTC 2020
This seems completely impractical/unworkable to me - who is going to be
maintaining this list of rules?
I see many issues on a first glance e.g.
- the hotspot interpreter belongs to runtime not hotspot compiler and
certainly not compiler-dev which is for javac!
- Anything under "make" has to go to build-dev as the primary list, as
well as, potentially, whatever component area it relates to.
- anything under test/hotspot goes to a hotspot mailing list not e.g.
On 27/08/2020 8:34 pm, Robin Westberg wrote:
> Hi all,
> As part of transitioning the jdk/jdk repository to Git with project Skara, we have created a set of rules that determine which mailing list(s) should be the default recipient of a review request, depending on which files are changed. The initial version of these rules was created by looking at historical commits and matching them with existing mailing list review threads. This has produced a reasonable basis, but it can most certainly be made better with some additional manual curating.
> Therefore, it would be very helpful if people with good knowledge of the various subsystems and source files that make up the JDK would be willing to take a look at these rules, and also suggest improvements where needed. In addition, lists like  would also be very useful insofar they exist.
> The current version of these rules is located in a JSON file in the Skara repository at . In order to check the validity of the rules, there is also a CLI tool that can be used to apply it to either a subset of files or existing commits and produce a suggestion  . If you are interested in helping out with curating these rules, these are the steps to get started:
> 1. Install the Skara CLI tools: https://wiki.openjdk.java.net/display/SKARA/CLI+Tools
> 2. Locate a suitable clone of the jdk/jdk repository (either Mercurial or Git is fine)
> 3. Change (cd) to the root of your jdk/jdk repository
> 3. Run the “debug mlrules” command on your favorite subset of files, for example like this (use the actual location of jdk.json on your system):
> $ git skara debug mlrules -v ~/git/skara/config/mailinglist/rules/jdk.json src/hotspot/share/jfr/
> Reading rules file...
> src/hotspot/share/jfr/dcmd: [hotspot-jfr-dev]
> src/hotspot/share/jfr/instrumentation/jfrEventClassTransformer.cpp: [hotspot-jfr-dev]
> src/hotspot/share/jfr/instrumentation/jfrEventClassTransformer.hpp: [hotspot-jfr-dev]
> src/hotspot/share/jfr/instrumentation/jfrJvmtiAgent.cpp: [hotspot-jfr-dev, serviceability-dev]
> Final list suggestion is: [hotspot-jfr-dev, serviceability-dev]
> The command accepts multiple folder and/or file names to make it possible to simulate a potential change to a given set of files:
> $ git skara debug mlrules -v ../skara/config/mailinglist/rules/jdk.json doc/ide.md src/hotspot/cpu/x86/x86.ad src/hotspot/os/linux/gc/z/zNUMA_linux.cpp
> Reading rules file...
> doc: [build-dev]
> src/hotspot/cpu: [hotspot-compiler-dev]
> src/hotspot/os: [hotspot-runtime-dev, hotspot-gc-dev]
> Combined list suggestion: [build-dev, hotspot-compiler-dev, hotspot-gc-dev, hotspot-runtime-dev]
> Final list suggestion is: [build-dev, hotspot-dev]
> If the suggestions look fine, all is well. If not, you are welcome to propose a change to the rules, preferably by editing the jdk.json file  and creating a pull request towards the Skara project as described in . Coincidentally, this is the same way that future changes to the jdk/jdk repository will be integrated, so this exercise could also serve as a way of getting started with Git / Skara!
> Best regards,
>  https://openjdk.java.net/groups/2d/2dawtfiles.html
>  https://git.openjdk.java.net/skara/blob/master/config/mailinglist/rules/jdk.json
> $ git skara debug mlrules -v ~/git/skara/config/mailinglist/rules/jdk.json src/java.desktop/unix/native
> Reading rules file...
> src/java.desktop/unix/native/common: [2d-dev]
> src/java.desktop/unix/native/include: 
> src/java.desktop/unix/native/libawt: [2d-dev]
> src/java.desktop/unix/native/libawt_headless: [awt-dev]
> src/java.desktop/unix/native/libawt_xawt: [awt-dev]
> src/java.desktop/unix/native/libfontmanager: [2d-dev]
> src/java.desktop/unix/native/libjawt: [awt-dev]
> src/java.desktop/unix/native/libsplashscreen: [awt-dev]
> Combined list suggestion: [2d-dev, awt-dev]
> Final list suggestion is: [2d-dev, awt-dev]
> $ git skara debug mlrules -d 30 -v ~/git/skara/config/mailinglist/rules/jdk.json .
> ✅ [2d-dev, awt-dev, serviceability-dev] c32923e0: 8240487: Cleanup whitespace in .cc, .hh, .m, and .mm files
> ❌ [awt-dev] 7f74c7dd: 8212226: SurfaceManager throws "Invalid Image variant" for MultiResolutionImage (Windows)
> Suggested lists: [2d-dev, awt-dev]
> Rules matching unmentioned lists [2d-dev]:
> src/java.desktop/share/classes/sun/java2d/SunGraphics2D.java - [2d-dev: ^src/java.desktop/share/classes/sun/java2d/]
>  https://wiki.openjdk.java.net/display/SKARA/Skara#Skara-Workflow
> The rules are made up of sets of regular expressions for the various mailing lists that are used for reviewing changes going into the JDK. If any rule matches, that mailing list will get a copy of the review request email. For directories containing files that belong to different subsystems, it’s usually a good idea to write the rules in a complementary fashion if possible, so that anything not explicitly mentioned gets a reasonable default. As an example, see these rules for a subset of awt / 2d / i18n files:
> “awt-dev”: "src/java.desktop/share/classes/sun/awt/(?!font|sunhints|color/|font/|geom/|im/|image/|print/)”
> “2d-dev”: "src/java.desktop/share/classes/sun/awt/(font|sunhints|color/|font/|geom/|image/|print/)"
> “I18n-dev”: "src/java.desktop/share/classes/sun/awt/im/“
> In this example, anything not explicitly indicated as belonging to either 2d-dev or i18-dev will be matched by awt-dev.
More information about the jdk-dev