Skip to content

Latest commit

 

History

History
188 lines (158 loc) · 7.59 KB

pipeline.md

File metadata and controls

188 lines (158 loc) · 7.59 KB

From the pipeline to the UCD

The following checklist for preparing a pull request with the UCD changes for an encoding proposal was (mostly) followed for https://github.com/unicode-org/unicodetools/pulls?q=label%3Apipeline-16.0. The plan is for this process to be part of the PAG’s review of encoding proposals going forward.

Checklist

Prerequisites: proposal posted to L2, SAH agreed to recommend for provisional assignment (or the proposal is already in the pipeline).

  • UnicodeData.txt — Prepend lines from proposal
  • Commit
  • UTC decision — Check counts, code points, names, properties
  • SAH report — Check counts, code points, names, properties
  • Ken’s UnicodeData draft — Check consistent

If the proposal supplies LineBreak.txt:

  • LineBreak.txt — Prepend lines from proposal
  • Commit

If the proposal does not supply LineBreak.txt:

  • LineBreak.txt — Regenerate [TODO(markus): This should become « invoke Ken’s tool »]
  • Update modified lines
  • Commit

New scripts only:

  • UCD_Names — Check script name
  • Scripts.txt — Prepend ranges (carefully mind any gaps)
  • Commit

New blocks only:

  • ShortBlockNames.txt — Update, keep sorted
  • Blocks.txt — Update, keep sorted [TODO(egg): This one wants to be generated…]
  • Commit

Joining scripts only:

  • ArabicShaping.txt — Merge from proposal, keep sorted
  • Commit

Indic scripts only:

  • IndicPositionalCategory — Prepend lines from proposal
  • IndicSyllabicCategory — Prepend lines from proposal
  • Commit
  • PropsList.txt — Add Other_Alphabetic, Other_Lowercase, Diacritic, and Extender to satisfy invariants, or to taste
  • Commit

PR preparation:

  • If from SAH — Link SAH issue
  • If from ESC or CJK — Mention ESC or CJK in the PR description
  • When for a UTC decision — Cite in the format UTC-\d\d\d-[MC]\d+ or with a link.
  • Whenever there is a Proposal document — Cite L2 number in the format L2/yy-nnn
  • data-for-new — Set label
  • pipeline-* — Set label to pipeline-recommended-to-UTC if the characters are not yet in the pipeline, and pipeline-provisionally-assigned, or pipeline-<version> depending on their status in the Pipeline.
  • PR button — Set to DRAFT pull request
    • unless approved for the upcoming version
  • PR button — Press
    • The Check UCA data and Check security data invariants CI checks are suppressed; many character additions need separate handling there, but that is out of scope for the PAG work of preparing data-for-new, so reporting those failures could distract from real issues in the UCD invariants. UCA and security data issues are addressed later in the process, before the start of β review.

Scripts

There are a variety of setups for unicodetools, depending on OS, in-source vs. out-of-source, git practices, etc. If you take part in UCD development, feel free to add your own.

Ken UnicodeData

Ken's files come from here (select appropriate ucd version e.g. ucd160 for Unicode 16.0). NOTE: this check is probably not applicable for pipeline-provisionally-assigned data where Ken does not yet have a draft.

eggrobin (Windows, in-source; the remote corresponding to unicode-org is called la-vache, Ken’s files are downloaded next to the unicodetools repository).

$latestKenFile = (ls ..\UnicodeData-*.txt | sort LastWriteTime)[-1]
$kenUnicodeData = (Get-Content $latestKenFile)
git diff la-vache/main */UnicodeData.txt |
sls ^\+[0-9A-F]                          |
% {
  $headLine = $_.line.Substring(1)
  if (-not $kenUnicodeData.Contains($headLine)) {
    $codepoint = $headLine.Split(";")[0];
    echo "Mismatch for U+$codepoint";
    echo "HEAD : $headLine";
    echo "Ken  : $($kenUnicodeData.Where({$_.Split(";")[0] -eq $codepoint}))";
  }
}

Merge

eggrobin (Windows, in-source; the remote corresponding to unicode-org is called la-vache).

git fetch la-vache
git merge la-vache/main
git checkout la-vache/main unicodetools/data/ucd/dev/Derived*;
git checkout la-vache/main unicodetools/data/ucd/dev/extracted/*;
git checkout la-vache/main unicodetools/data/ucd/dev/auxiliary/*;
rm .\Generated\* -recurse -force;
mvn compile exec:java '-Dexec.mainClass="org.unicode.text.UCD.Main"'  '-Dexec.args="build MakeUnicodeFiles"' -am -pl unicodetools  "-DCLDR_DIR=..\cldr\"  "-DUNICODETOOLS_GEN_DIR=Generated"  "-DUNICODETOOLS_REPO_DIR=.";
cp .\Generated\UCD\16.0.0\* .\unicodetools\data\ucd\dev -recurse -force;
rm unicodetools\data\ucd\dev\zzz-unchanged-*;
rm unicodetools\data\ucd\dev\*\zzz-unchanged-*;
rm .\unicodetools\data\ucd\dev\extra\*;
rm .\unicodetools\data\ucd\dev\cldr\*;
git add ./unicodetools/data
git merge --continue

markusicu (Linux, out-of-source; main tracks unicode-org/main)

git merge main
# complains about merge conflicts as expected
git checkout main unicodetools/data/ucd/dev/Derived*
git checkout main unicodetools/data/ucd/dev/extracted/*
git checkout main unicodetools/data/ucd/dev/auxiliary/*
rm -r ../Generated/BIN/16.0.0.0/
rm -r ../Generated/BIN/UCD_Data16.0.0.bin
mvn -s ~/.m2/settings.xml compile exec:java -Dexec.mainClass="org.unicode.text.UCD.Main"  -Dexec.args="version 16.0.0 build MakeUnicodeFiles" -am -pl unicodetools  -DCLDR_DIR=$(cd ../../../cldr/mine/src ; pwd)  -DUNICODETOOLS_GEN_DIR=$(cd ../Generated ; pwd)  -DUNICODETOOLS_REPO_DIR=$(pwd)  -DUVERSION=16.0.0
# fix merge conflicts in unicodetools/src/main/java/org/unicode/text/UCD/UCD_Types.java
#   and in UCD_Names.java
# rerun mvn
cp -r ../Generated/UCD/16.0.0/* unicodetools/data/ucd/dev
rm unicodetools/data/ucd/dev/ZZZ-UNCHANGED-*
rm unicodetools/data/ucd/dev/*/ZZZ-UNCHANGED-*
rm unicodetools/data/ucd/dev/extra/*
rm unicodetools/data/ucd/dev/cldr/*
git add unicodetools/src/main/java/org/unicode/text/UCD/UCD_Names.java
git add unicodetools/src/main/java/org/unicode/text/UCD/UCD_Types.java
git add unicodetools/data
git merge --continue

macchiati (IDE)

sync github
run MakeUnicodeFiles.java -c

Cf. unicode-org#636

Regenerate UCD

eggrobin (Windows, in-source).

rm .\Generated\* -recurse -force
mvn compile exec:java '-Dexec.mainClass="org.unicode.text.UCD.Main"'  '-Dexec.args="build MakeUnicodeFiles"' -am -pl unicodetools  "-DCLDR_DIR=..\cldr\"  "-DUNICODETOOLS_GEN_DIR=Generated"  "-DUNICODETOOLS_REPO_DIR=."
cp .\Generated\UCD\16.0.0\* .\unicodetools\data\ucd\dev -recurse -force
rm unicodetools\data\ucd\dev\zzz-unchanged-*
rm unicodetools\data\ucd\dev\*\zzz-unchanged-*
rm .\unicodetools\data\ucd\dev\extra\*
rm .\unicodetools\data\ucd\dev\cldr\*
git add unicodetools/data/ucd/dev/*
git commit -m "Regenerate UCD"

Regenerate LineBreak

eggrobin (Windows, in-source).

rm .\Generated\* -recurse -force
mvn compile exec:java '-Dexec.mainClass="org.unicode.text.UCD.Main"'  '-Dexec.args="build MakeUnicodeFiles"' -am -pl unicodetools  "-DCLDR_DIR=..\cldr\"  "-DUNICODETOOLS_GEN_DIR=Generated"  "-DUNICODETOOLS_REPO_DIR=."
cp .\Generated\UCD\16.0.0\LineBreak.txt .\unicodetools\data\ucd\dev

GenerateEnums

eggrobin (Windows, in-source).

mvn compile exec:java '-Dexec.mainClass="org.unicode.props.GenerateEnums"' -am -pl unicodetools  "-DCLDR_DIR=..\cldr\"  "-DUNICODETOOLS_GEN_DIR=Generated"  "-DUNICODETOOLS_REPO_DIR=." -U
mvn spotless:apply
git add *.java
git commit -m GenerateEnums