Bug #17747: `make up` is broken with clean repo.

• Status: Closed
• Assignee: -
• Target version: -
• Backport: 2.5: UNKNOWN, 2.6: UNKNOWN, 2.7: UNKNOWN, 3.0: UNKNOWN

[hsbt (Hiroshi SHIBATA)]

make up is broken when after cloning git repo like https://github.com/ruby/ruby.

expected instructions

./autogen.sh
./configure # with some options
make ruby -j
make up

it's failed with checking unicode file existence.

$ make up
remote: Counting objects: 3, done.
remote: Compressing objects: 100% (3/3), done.
remote: Total 3 (delta 2), reused 0 (delta 0)
Unpacking objects: 100% (3/3), 1.03 KiB | 29.00 KiB/s, done.
From git.ruby-lang.org:ruby
   b25361f731..12812c2bdf  master     -> origin/master
Updating b25361f731..12812c2bdf
Fast-forward
 common.mk | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)
Latest commit hash = 12812c2bdf
tool/config.guess already exists
tool/config.sub already exists
Downloading bundled gem minitest-5.14.4...
Downloading bundled gem power_assert-2.0.0...
Downloading bundled gem rake-13.0.3...
Downloading bundled gem test-unit-3.4.0...
Downloading bundled gem rexml-3.2.4...
Downloading bundled gem rss-0.2.9...
Downloading bundled gem typeprof-0.13.0...
Downloading bundled gem rbs-1.1.1...
downloading for ext/fiddle/extlibs
downloading libffi-3.2.1.tar.gz ... done
checking md5 of .downloaded-cache/libffi-3.2.1.tar.gz ... OK
checking sha512 of .downloaded-cache/libffi-3.2.1.tar.gz ... OK
downloading for ext/zlib/extlibs
downloading zlib-1.2.11.tar.gz ... done
checking md5 of .downloaded-cache/zlib-1.2.11.tar.gz ... OK
checking sha512 of .downloaded-cache/zlib-1.2.11.tar.gz ... OK
make[1]: Entering directory '/Users/hsbt/Documents/github.com/ruby/ruby'
sed 's/{\$([^(){}]*)[^{}]*}//g' common.mk > uncommon.mk
./revision.h unchanged
making gem_prelude.rbinc
make[2]: Entering directory '/Users/hsbt/Documents/github.com/ruby/ruby'
make[2]: Leaving directory '/Users/hsbt/Documents/github.com/ruby/ruby'
/Users/hsbt/Documents/github.com/ruby/ruby/tool/lib/vpath.rb:18:in `initialize': No such file or directory @ rb_sysopen - enc/unicode/data/12.1.0/ucd/CompositionExclusions.txt (Errno::ENOENT)
        from /Users/hsbt/Documents/github.com/ruby/ruby/tool/lib/vpath.rb:18:in `open'
        from /Users/hsbt/Documents/github.com/ruby/ruby/tool/lib/vpath.rb:18:in `call'
        from /Users/hsbt/Documents/github.com/ruby/ruby/tool/lib/vpath.rb:18:in `search'
        from /Users/hsbt/Documents/github.com/ruby/ruby/tool/lib/vpath.rb:35:in `open'
        from ./template/unicode_norm_gen.tmpl:69:in `block (2 levels) in <main>'
        from /Users/hsbt/.local/share/rbenv/versions/3.0.0-dev/lib/ruby/3.0.0/erb.rb:905:in `eval'
        from /Users/hsbt/.local/share/rbenv/versions/3.0.0-dev/lib/ruby/3.0.0/erb.rb:905:in `result'
        from ./tool/generic_erb.rb:41:in `block (2 levels) in <main>'
        from ./tool/generic_erb.rb:41:in `block in <main>'
        from ./tool/generic_erb.rb:34:in `map'
        from ./tool/generic_erb.rb:34:in `<main>'
make[1]: *** [uncommon.mk:1491: enc/unicode/data/12.1.0/ucd/.unicode-tables.time] Error 1
make[1]: Leaving directory '/Users/hsbt/Documents/github.com/ruby/ruby'
make: [uncommon.mk:1283: up] Error 2 (ignored)

The current implementation works fine to run make update-unicode before make up

I think it's not intentional behavior.

[hsbt (Hiroshi SHIBATA)]

• Description updated (diff)

[hsbt (Hiroshi SHIBATA)]

• Description updated (diff)

[hsbt (Hiroshi SHIBATA)]

I confirmed to work with https://github.com/ruby/ruby/commit/041a4f9a5e04eb4e6f8f26d143a613a817e18717

But, the current make up didn't invoke make update-unicode. I'm not sure it's required with make up.

@nobu (Nobuyoshi Nakada) Can you explain that?

[nobu (Nobuyoshi Nakada)]

The files generated from Unicode data files (enc/unicode/12.1.0/{casefold,name2ctype}.h and lib/unicode_normalize/tables.rb) are committed, and Unicode data is (except for an exception) updated only once per year now.

Also, CIs were banned from Unicode.org when they were downloading the data every time.

[hsbt (Hiroshi SHIBATA)]

• Status changed from Open to Closed

Thanks for explaining. It's reasonable.