lib/rdoc/generator/json_index.rb | 4 2 + 2 - 0 !
1 file changed, 2 insertions(+), 2 deletions(-)

 rdoc: build reproducible documentation

- provide a fixed timestamp to the gzip compression

Signed-off-by: Antonio Terceiro <terceiro@debian.org>
Signed-off-by: Christian Hofstaedtler <zeha@debian.org>

lib/mkmf.rb | 2 1 + 1 - 0 !
1 file changed, 1 insertion(+), 1 deletion(-)

 lib/mkmf.rb: sort list of object files in generated makefile

Without sorting the list explicitly, its order is indeterministic,
because readdir() is also not deterministic.
When the list of object files varies between builds, they are linked

lib/rubygems/specification.rb | 4 3 + 1 - 0 !
1 file changed, 3 insertions(+), 1 deletion(-)

 mark gemspec-reproducible change fixing #784225, too

I think the UTC date change will fix the Multi-Arch not-same file issue,
too.

Signed-off-by: Antonio Terceiro <terceiro@debian.org>
Signed-off-by: Christian Hofstaedtler <zeha@debian.org>

test/.excludes/TestFileUtils.rb | 1 1 + 0 - 0 !
test/.excludes/TestProcess.rb | 1 1 + 0 - 0 !
2 files changed, 2 insertions(+)

 disable tests failing on ubuntu builders

.bundle/gems/bigdecimal-3.1.8/bigdecimal.gemspec | 1 1 + 0 - 0 !
ext/fiddle/fiddle.gemspec | 1 1 + 0 - 0 !
ext/io/console/io-console.gemspec | 1 1 + 0 - 0 !
lib/ipaddr.gemspec | 1 1 + 0 - 0 !
lib/rdoc/rdoc.gemspec | 1 1 + 0 - 0 !
5 files changed, 5 insertions(+)

 make gemspecs reproducible

Without an explicit date, they will get the current date and make the
build unreproducible

io.c | 6 5 + 1 - 0 !
1 file changed, 5 insertions(+), 1 deletion(-)

 fix ftbs on hurd

Closes: #896509

Signed-off-by: Antonio Terceiro <terceiro@debian.org>
Signed-off-by: Samuel Thibault <samuel.thibault@ens-lyon.org>

ext/socket/option.c | 1 1 + 0 - 0 !
ext/socket/raddrinfo.c | 22 22 + 0 - 0 !
test/fiddle/test_handle.rb | 1 1 + 0 - 0 !
test/socket/test_socket.rb | 2 1 + 1 - 0 !
4 files changed, 25 insertions(+), 1 deletion(-)

 port to kfreebsd-amd64

Closes: #899267

Signed-off-by: Antonio Terceiro <terceiro@debian.org>
Signed-off-by: Svante Signell <svante.signell@gmail.com>

configure.ac | 3 3 + 0 - 0 !
1 file changed, 3 insertions(+)

 fix ftbfs on x32: misdetected as i386 or amd64

Bug: #954293

defs/gmake.mk | 4 1 + 3 - 0 !
1 file changed, 1 insertion(+), 3 deletions(-)

 don't try to download stuff in configure

tool/m4/ruby_default_arch.m4 | 1 1 + 0 - 0 !
1 file changed, 1 insertion(+)

 ruby_default_arch.m4: don't require arhitectures to be known ahead
 of time

Not sure yet if this should be forwarded upstream or not.

tool/m4/ruby_default_arch.m4 | 1 0 + 1 - 0 !
1 file changed, 1 deletion(-)

 do not pass -m32 to arm 32 bits
MIME-Version: 1.0
Content-Type: text/plain; charset="utf-8"
Content-Transfer-Encoding: 8bit

The -m32 flag in already 32 bits architecture is generating the
following error:

arm-linux-gnueabi-gcc: error: unrecognized command-line option -m32; did you mean -mbe32?

test/did_you_mean/helper.rb | 28 16 + 12 - 0 !
test/rdoc/rdoc_markdown_test.rb | 9 7 + 2 - 0 !
test/rdoc/rdoc_markdown_test_test.rb | 9 7 + 2 - 0 !
test/rdoc/support/test_case.rb | 6 5 + 1 - 0 !
4 files changed, 35 insertions(+), 17 deletions(-)

 fix require calls falling in autopkgtest

ext/socket/lib/socket.rb | 6 6 + 0 - 0 !
1 file changed, 6 insertions(+)

 deduplicate results from getaddrinfo

In certain configurations it's possible to get duplicate results back
from getaddrinfo: for example, if you accidentally have more than one
line in /etc/hosts mapping the same name to the same IP address, then
Linux/glibc systems will return multiple identical entries.  This minor
misconfiguration is normally harmless, but it caused this program to
fail with `EADDRINUSE`:

    require 'socket'
    Socket.tcp_server_sockets('localhost', 0)

Prior to https://github.com/ruby/net-http/pull/180, this caused a number
of `TestNetHTTP*` tests to fail, as seen in these Debian bugs:

  https://bugs.debian.org/1069399 (Ruby 3.1)
  https://bugs.debian.org/1064685 (Ruby 3.2)
  https://bugs.debian.org/1077462 (Ruby 3.3)

It's easy enough to deduplicate these.

test/rubygems/test_bundled_ca.rb | 1 1 + 0 - 0 !
1 file changed, 1 insertion(+)

 [patch] test: test_bundled_ca: skip test because it requires
 internet access.

Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>

test/fiddle/test_fiddle.rb | 2 1 + 1 - 0 !
1 file changed, 1 insertion(+), 1 deletion(-)

 fix test dlopen for .so filename with more than one digit

In alpha architecture, we have libc.so.6.1 which causes a failure in
this test:

 1) Failure:
TestFiddle#test_dlopen_linker_script_group_linux [/<<PKGBUILDDIR>>/test/fiddle/test_fiddle.rb:52]:
<"libc.so"> expected but was
<"libc.so.6">.

Bug-Debian: https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=1085238

lib/rdoc/rdoc.gemspec | 3 2 + 1 - 0 !
1 file changed, 2 insertions(+), 1 deletion(-)

 rdoc: install gemspec and change lib_file dir.glob expression

test/reline/test_reline.rb | 2 1 + 1 - 0 !
1 file changed, 1 insertion(+), 1 deletion(-)

 test_tty_ambiguous_width: use reline.test_rubybin

Same as https://github.com/ruby/reline/pull/510, 'ruby' command is not
always available so don't rely on that specific name.

tool/mkrunnable.rb | 11 2 + 9 - 0 !
tool/rbinstall.rb | 4 1 + 3 - 0 !
2 files changed, 3 insertions(+), 12 deletions(-)

 put ruby interpreter inside /usr/bin

This was the behaviour up to ruby3.3
revert partially https://github.com/ruby/ruby/pull/10010
see also https://bugs.ruby-lang.org/issues/20800

tool/rbinstall.rb | 5 0 + 5 - 0 !
1 file changed, 5 deletions(-)

 no symlink to pkgconfig files in non arch-specific location