April 2024 - jextract-dev

RFR: 7903720: Change layout of jextract image
by Jorn Vernee 30 Apr '24

30 Apr '24

See JBS issue for problem description. This patch changes the layout of the jextract image we produce. The new image has the 2 launcher scripts in the root of the image, with a nested `runtime` folder containing the jlink runtime image. I've added both launcher scripts to the repo under the `src/main` directory, and they are copied to the image during the build. The script is somewhat simplified by removing a blank variable that was not being used. I also took the opportunity to address an issue with the windows launcher script turning off `echo`. Instead it now avoids output using the `@` prefix for each command in the script. I've tested that the scripts work (using `jextract --help`) on both Linux and Windows. ------------- Commit messages: - remove spurious imports - update verify task - Improve image layout and launcher scripts Changes: https://git.openjdk.org/jextract/pull/241/files Webrev: https://webrevs.openjdk.org/?repo=jextract&pr=241&range=00 Issue: https://bugs.openjdk.org/browse/CODETOOLS-7903720 Stats: 18 lines in 4 files changed: 12 ins; 1 del; 5 mod Patch: https://git.openjdk.org/jextract/pull/241.diff Fetch: git fetch https://git.openjdk.org/jextract.git pull/241/head:pull/241 PR: https://git.openjdk.org/jextract/pull/241

2 9

jextract wayland-client
by Thiago Milczarek Sayão 21 Apr '24

21 Apr '24

Hi, I'm trying to make a java wayland client (currently as a proof of concept). I've attempted to generate the bindings like this: jextract --output src -t org.freedesktop.wayland.client --header-class-name WlClient -lwayland-client /usr/include/wayland-client.h But the wl_xxx types seem to not be supported as it outputs: WARNING: Skipping wl_registry (type Declared(wl_registry) is not supported) I've probably hit a current limitation. Is it expected to work in the future? Thanks. -- Thiago.

2 5

RFR: Fix actions after version bump
by Jorn Vernee 18 Apr '24

18 Apr '24

JDK bundles on different platforms have a different internal directory structure. We currently hard-code the path to the 'root' of the JDK (containing the `bin`/`lib`/`include` dirs, etc.) for each platform. Recently the JDK version was bumped from 22 to 22.0.1, which changed the internal structure (since it includes the version in one of the directory names), thereby breaking our use of hard-coded paths. This PR intends to fix this issue once and for all, by getting rid of the hard-coded paths. Instead, we try to find the `bin` folder within the downloaded package, and then copy the files from this discovered JDK root to a known location with a canonical path that is similar enough on each platform (this trick is borrowed from the mainline JDK GHA). ------------- Commit messages: - Canonicallize toolchain structure Changes: https://git.openjdk.org/jextract/pull/239/files Webrev: https://webrevs.openjdk.org/?repo=jextract&pr=239&range=00 Stats: 20 lines in 1 file changed: 8 ins; 3 del; 9 mod Patch: https://git.openjdk.org/jextract/pull/239.diff Fetch: git fetch https://git.openjdk.org/jextract.git pull/239/head:pull/239 PR: https://git.openjdk.org/jextract/pull/239

2 2

RFR: Add guide section about platform-dependent nature of jextract
by Jorn Vernee 18 Apr '24

18 Apr '24

The current guide is lacking a discussion about the platform dependent nature of jextract, leaving users to wonder if and when the bindings generated by jextract can be used on different platforms. This PR adds a couple of paragraphs to the guide to explain this. Note: I've also fixed a typo in the README. ------------- Commit messages: - use italic - phrasing - Add notes on platform-dependent nature of jextract Changes: https://git.openjdk.org/jextract/pull/237/files Webrev: https://webrevs.openjdk.org/?repo=jextract&pr=237&range=00 Stats: 22 lines in 2 files changed: 21 ins; 0 del; 1 mod Patch: https://git.openjdk.org/jextract/pull/237.diff Fetch: git fetch https://git.openjdk.org/jextract.git pull/237/head:pull/237 PR: https://git.openjdk.org/jextract/pull/237

2 8

RFR: 7903715: Jextract generates duplicate allocator parameters
by Maurizio Cimadamore 18 Apr '24

18 Apr '24

This simple fix addresses a case where jextract generates two function parameter with duplicate name `allocator`. ------------- Commit messages: - Initial push Changes: https://git.openjdk.org/jextract/pull/238/files Webrev: https://webrevs.openjdk.org/?repo=jextract&pr=238&range=00 Issue: https://bugs.openjdk.org/browse/CODETOOLS-7903715 Stats: 75 lines in 3 files changed: 71 ins; 4 del; 0 mod Patch: https://git.openjdk.org/jextract/pull/238.diff Fetch: git fetch https://git.openjdk.org/jextract.git pull/238/head:pull/238 PR: https://git.openjdk.org/jextract/pull/238

2 3

(no subject)
by Paulo Feodrippe 18 Apr '24

18 Apr '24

Hi, I'm using Build 22-jextract+3-13 and, while I'm able to jextract the flecs.c file in https://github.com/SanderMertens/flecs, it generates a param with the same name as one of the original params (allocator). ./src-java/org/vybe/flecs_2.java:7946: error: variable allocator is already defined in method ecs_vec_copy(java.lang.foreign.SegmentAllocator,java.lang.foreign.MemorySegment,java.lang.foreign.MemorySegment,int) public static MemorySegment ecs_vec_copy(SegmentAllocator allocator, MemorySegment allocator, MemorySegment vec, int size) { ^ ./src-java/org/vybe/flecs_2.java:7995: error: variable allocator is already defined in method ecs_vec_copy_shrink(java.lang.foreign.SegmentAllocator,java.lang.foreign.MemorySegment,java.lang.foreign.MemorySegment,int) public static MemorySegment ecs_vec_copy_shrink(SegmentAllocator allocator, MemorySegment allocator, MemorySegment vec, int size) { Thanks -- Paulo Rafael Feodrippe, Desenvolvedor

2 2

New guide --include-dir clarifications
by Nir Lisker 17 Apr '24

17 Apr '24

Hi, A couple of points regarding --include-dir in the guide that I think are worth mentioning: 1. The directories are searched in order, so if the same header is found in more than one included dirs, only the first one is taken. (I think this is correct.) 2. The directories are *not* searched recursively. Nested directories need to be added one by one if required. If you agree that these should be added, I don't mind making a pull request. I think it's best to add this info in the "Running jextract" section. - Nir

3 2

RFR: Add guide section about array types
by Jorn Vernee 16 Apr '24

16 Apr '24

The current guide is missing a section discussing the special handling of array types, for which jextract generates a `$dimensions()` accessor, and indexed getters and setters. This PR adds a section to the guide discussing this. I've put it at the end of the "Using The Code Generated By Jextract" section. I've put it before the section on nested types though, since it's a little more important I think. ------------- Commit messages: - Add guide section about array types Changes: https://git.openjdk.org/jextract/pull/236/files Webrev: https://webrevs.openjdk.org/?repo=jextract&pr=236&range=00 Stats: 61 lines in 1 file changed: 61 ins; 0 del; 0 mod Patch: https://git.openjdk.org/jextract/pull/236.diff Fetch: git fetch https://git.openjdk.org/jextract.git pull/236/head:pull/236 PR: https://git.openjdk.org/jextract/pull/236

2 9

RFR: Add jextract guide
by Jorn Vernee 16 Apr '24

16 Apr '24

Add a comprehensive jextract guide under a new `doc/` folder. This is meant as a comprehensive guide about the features of jextract and the code that it generates (including both examples from header files, the code that jextract generates, and corresponding Java user code). This is a first cut, and I'm anticipating quite a bit of comments. Some sections of the readme have been moved to the guide (with minor edits here and there), and replaced by a link in the readme. ------------- Commit messages: - Phrasing - Section of pre-precessor defines + other languages - Polish - Update README.md - Create GUIDE.md Changes: https://git.openjdk.org/jextract/pull/231/files Webrev: https://webrevs.openjdk.org/?repo=jextract&pr=231&range=00 Stats: 995 lines in 2 files changed: 824 ins; 171 del; 0 mod Patch: https://git.openjdk.org/jextract/pull/231.diff Fetch: git fetch https://git.openjdk.org/jextract.git pull/231/head:pull/231 PR: https://git.openjdk.org/jextract/pull/231

3 89

RFR: Add Windows support in GitHub actions
by Jorn Vernee 12 Apr '24

12 Apr '24

We currently don't run test in GitHub action on Windows. This can mean that a build failure on Windows may be missed (as seen recently). This PR adds support for running tests on Windows to the GitHub actions workflow. Some notes about the changes: - We can not build jtreg on Windows, so I've added a separate job which builds it on Ubuntu and then caches it. Then the main test job restores it from this cache. (FWIW, this cache is private/can not be downloaded outside of the actions) - The `oracle-actions/setup-java` action we use gives us back a Windows path on Windows, which I don't think we can easily handle in bash, so I've created a Windows specific step for extracting the downloaded toolchain JDK - Added caches also for LLVM/libclang and the toolchain JDK, so we don't have to download them all the time, and testing is faster. - For that reason, I've added a step that creates a trimmed-down 'image' of the LLVM package with only the bits that we need, so that we can only cache those bits, instead of the entire LLVM package. ------------- Commit messages: - use cache to restore jtreg in main job - rename step - Don'trebuild jtreg on cache hit - Trim down LLVM package for cache - cache jtreg too - remove actions files again - Try to use separate action scripts - Try add Windows support to GHA Changes: https://git.openjdk.org/jextract/pull/235/files Webrev: https://webrevs.openjdk.org/?repo=jextract&pr=235&range=00 Stats: 116 lines in 1 file changed: 78 ins; 7 del; 31 mod Patch: https://git.openjdk.org/jextract/pull/235.diff Fetch: git fetch https://git.openjdk.org/jextract.git pull/235/head:pull/235 PR: https://git.openjdk.org/jextract/pull/235

2 2