Repository navigation
docs: move the unreleased changes under 0.6.5, check the dated claims again, correct the Corretto checksum remarks and note that Foojay has no CRaC builds - #125
Merged
Conversation
… and correct what the docs say about the checksum of Corretto After the release: - `docs/changes.md` and its translation: "Unreleased" becomes 0.6.5 (2026-11-03), and the sentence about `main` being ahead of the latest release goes. - `docs/sdkmanrc.md` and its translation: the "Unreleased: release 0.6.4 ..." remark goes. `changes.md` has what it said. Corrected, found while comparing the quoted data with Foojay: - `docs/foojay.md`, `docs/troubleshooting.md`, the 0.6.4 entry of `docs/changes.md` and their translations said that Corretto 8 comes with a checksum file and that only the releases after 8 do not. Foojay lists an MD5 file only on a record without a build number (`8`, and likewise 11 to 25), which the plugin never installs. `java@8-amzn` installs `8.0.512-amzn`, like every other Corretto install, and vfox warns that there is no checksum. - The checksum table and its example in `docs/foojay.md` were compared again, so their date is 2026-11-03.
…eck of e2e.md again, and say that Foojay has no CRaC builds - `docs/native-image.md` and its translation: the checks dated 2026-11-02 were run again on 2026-11-03 in Ubuntu 26.04 and Fedora 44 containers, and every statement held. The "What was checked" paragraph now also names the other GraalVM CE releases: `22.3.3-grl11`, `21.2-grl16` and `22.3.1-grl19` after `gu install native-image`, and `21.3.1-grl8` after `gu -L install` of the downloaded file. - `docs/e2e.md` and its translation: Foojay lists a checksum for the pinned release of every default distribution on every runner, compared again on 2026-11-03 (`open` and `graalce` have no Windows ARM64 build, which the suite skips). - `docs/sdkmanrc.md` and its translation: SDKMAN lists CRaC builds of Liberica and Zulu (`21.0.13.0-crac+12-librca`), which it downloads from the vendors. None of Foojay's 35,562 JDK packages is a CRaC build, so the plugin has none to install and an identifier with `crac` selects nothing.
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Documentation only. After the release, as for 0.6.2 to 0.6.4:
docs/changes.mdand its translation: "Unreleased" becomes 0.6.5 (2026-11-03, the UTC day of the release), and the sentence aboutmainbeing ahead of the latest release goes.docs/sdkmanrc.mdand its translation: the "Unreleased: release 0.6.4 ..." remark about the number after the+in the identifiers of Oracle GraalVM and GraalVM Community goes.changes.mdhas what it said.grep -rn -i -e unreleased -e 未发布 docs README*now finds only the steps indevelopment.md(and "从未发布过" inzh-CN/e2e.md, which is no remark), no page names 0.6.4 outsidechanges.md, and no2026-11-02is left.Corrected, found while comparing the quoted data with Foojay:
docs/foojay.md,docs/troubleshooting.md, the 0.6.4 entry ofdocs/changes.mdand their translations said that Corretto 8 comes with a checksum file and that only "Corretto after 8" has none (fix: verify downloads against the checksum Foojay lists, sovfox install java@25.0.2-graalceand the like no longer end withWARNING: Checksum is not provided, skip verify...#120). Foojay lists an MD5 file only on a record without a build number (amazon-corretto-8-x64-linux-jdk.tar.gz,java_version8, and likewise 11 to 25), which the plugin never installs:java@8-amzninstalls8.0.512-amzn, and the numbered releases have no checksum.vfox plugin run . PreInstallgives no checksum for8-amzn,8.0.512-amzn,11-amzn,17-amzn,21-amzn,25-amznand27-amznon Linux x64, and for8-amznon Linux arm64, macOS x64 and arm64 and Windows x64 as well. So every Corretto install ends with vfox's warning, and the three pages now say that.Added:
docs/sdkmanrc.mdand its translation say that SDKMAN lists CRaC builds of Liberica and Zulu (21.0.13.0-crac+12-librca,25.0.4.1-crac+1-zulu), that it downloads them from the vendors (github.com/bell-sw/Liberica/...-crac.tar.gz,cdn.azul.com/zulu/bin/...-ca-crac-...), and that the plugin has none to install, so an identifier withcracselects nothing (ParseLegacyFilereturns an empty version). Foojay is not a source for them: none of the 35,562 JDK packages of its 31 distributions hascracin a field or a file name,feature=cracchanges no answer (thefeatureof the API is for early-access features of Oracle OpenJDK, see its spec), and foojayio/discoapi has no issue about it. Installing them would need a second source next to Foojay, as for Loongnix (BellSoft's and Azul's APIs); that is not part of this change.Dated claims run again on 2026-11-03, and their dates are 2026-11-03 now (they were 2026-11-02):
docs/native-image.mdand its translation, inubuntu:26.04containers without a compiler and infedora:44, with the plugin of this branch (vfox add java --source),vfox installandvfox use --global:native-imagestops after 1 to 6 s withDefault native-compiler executable 'gcc' not found via environment variable PATH, andzlib1g-devalone does not change that.gcc,libc6-devandzlib1g-dev, with nog++and nomake, build a hello world in 22 to 52 s and it runs.build-essentialwithoutzlib1g-devfails when it links, after 29 to 52 s, withLinker command exited with 1,It appears as though libz:.a is missingandcannot find -lz: No such file or directory. Withbuild-essentialandzlib1g-devit builds.22.3.3-grl11,21.2-grl16,22.3.3-grl17and22.3.1-grl19:native-imageiscommand not found,gu install native-imageworks (5 to 10 s), and the build then stops without the packages and builds with them.21.3.1-grl8:gu install native-imagefails withInternal error occurred: A JSONObject text must begin with '{', ashttps://oca.opensource.oracle.com/gds/meta-data.jsonanswers a web page (content-type: text/html), andgu -L installofnative-image-installable-svm-java8-linux-amd64-21.3.1.jarworks, after which the packages are needed as well.gcc,glibc-develandzlib-devel, withoutlibstdc++-static, build the hello world with GraalVM Community 25.0.2 in 35 s.22.3.3-grl17only.docs/e2e.mdand its translation: Foojay lists a checksum for the pinned release of every default distribution on every runner.vfox plugin run . PreInstallfor25.0.2(OpenJDK),21.0.8-tem,25.0.4-librca-fx,21.0.5-jband25.0.2-graalceon Linux and Windows, x64 and arm64, gives a SHA-256 (OpenJDK, Temurin, GraalVM Community), SHA-1 (Liberica) or SHA-512 (JetBrains) digest each time, as the paragraph says. OpenJDK and GraalVM Community have no Windows ARM64 build, which the suite skips.docs/foojay.md, the checksum table and its example: see the Corretto correction above.Data, compared on 2026-11-03 (queries as in
development.md, parsed with Python because the host has nojq):mirror.md: the newest build of every major and the oldest one, for Linux x64), the checksum table offoojay.mdfor the newest release of each distribution (apart from Corretto above), the numbers inREADME.md(11 distributions with Alpine x64, 7 with Alpine arm64, 5 with Windows ARM64), the notes under "Only some releases" for Windows ARM64, Apple silicon, Intel Macs, Alpine, Trava, AdoptOpenJDK's OpenJ9 and OJDKBuild, the examples with output infoojay.md, and the "Other CPUs" lists (32-bit ARM and x86, riscv64, ppc64le, s390x; Foojay still answers HTTP 400invalid architecturefor loong64).openjdk-sources.osci.io, which no longer has the files, and vfox ends withsource file not found), while 21.0.x for Linux x64 downloads fromdevelopers.redhat.com.java@21-temis21.0.13+9-tem,java@28.ea-openis28-ea+21,java@25-graalis25.0.5.0.3-graal,java@25-graalceis25.4.4.1.1-graalce(all asvfox plugin run . PreInstallanswers),vfox search java alllists 2,646 lines on Linux x64 ("about 2,600"), SDKMAN's default is25.0.5.0+8-tem, and its list has none of Dragonwell, Trava, AdoptOpenJDK, Gluon GraalVM, GraalVM CE 8 to 19, Eliya, OJDKBuild, OpenLogic, Red Hat and Semeru certified. Eliya's site still says its SDKMAN registration is in progress. vfox 1.0.13 is still the newest vfox.loongnix.md).The 0.6.5 entry is the plugin's behaviour: with
25.0.5-graal,25.0.5.0.3-graaland25.4.4.1.1-graalinstalled, a.sdkmanrcwithjava=25.0.5.0+3-graalselects25.0.5.0.3-graal, andjava=25.4.4.1+1-graalceselects25.4.4.1.1-graalce(vfox plugin run . ParseLegacyFile).vfox plugin test .passes, and the relative links and anchors of all 33 Markdown files resolve (340 links, a script that follows GitHub's rules for the anchors of headings; it was clean before the change too). The English and Chinese pages have the same headings, tables and code blocks apart from translated comments, as before the change.