Includes fixes for: CVE-2022-28739, CVE-2021-41816, and CVE-2021-41817.
* gnu/packages/ruby.scm (ruby-2.7-fixed): New variable.
(ruby-2.7)[replacement]: Graft.
Signed-off-by: Marius Bakke <marius@gnu.org>
Includes fixes for: CVE-2022-28738, and CVE-2022-28739.
* gnu/packages/ruby.scm (ruby-3.1): Update to 3.1.2.
Signed-off-by: Marius Bakke <marius@gnu.org>
Includes fixes for: CVE-2022-28738, CVE-2022-28739, CVE-2021-41819,
CVE-2021-41816, and CVE-2021-41817.
* gnu/packages/ruby.scm (ruby-3.0): Update to 3.0.4.
Signed-off-by: Marius Bakke <marius@gnu.org>
As of commit "gnu: unpin ruby-nokogiri-diff's ruby-nokogiri",
ruby-nokogiri-1.10 is unused, so let's remove it.
ruby-nokogiri-1.10's purpose was to cut the dependency graph. Users should use
ruby-nokogiri.
* gnu/packages/ruby.scm (ruby-nokogiri-1.10): Remove variable.
(ruby-nokogiri): Collapse the package inheritance.
Signed-off-by: Ludovic Courtès <ludo@gnu.org>
ruby-nokogiri-diff historically had a long reverse dependency set, due to its
reverse dependency of texlive, which was solely via ruby-byebugs.
To avoid ruby-nokogiri updates from triggering rebuilds, ruby-nokogiri-diff
was pinned to ruby-nokogiri-1.10.
However, as of commit "gnu: Remove texlive's dependence on ruby-rspec and
ruby-byebug.", texlive no longer depends on ruby-byebug, and thus updating
ruby-nokogiri won't trigger large rebuilds.
Thus, we are now free to unpin ruby-nokogiri-diff's ruby-nokogiri.
* gnu/packages/ruby.scm (ruby-nokogiri-diff): Unpin ruby-nokogiri version.
Signed-off-by: Ludovic Courtès <ludo@gnu.org>
Since we have GDBM available, it offers a smaller memory footprint
and faster start-up.
* gnu/packages/ruby.scm (anystyle)[arguments]<#:phases>: Add phase
'change-default-dictionary-adapter'.
Signed-off-by: Ludovic Courtès <ludo@gnu.org>
* gnu/packages/ruby.scm (anystyle)[arguments]: Instead of deleting the
'check' phase, replace it with a few tests. Add SRFI 1 to '#:modules'.
Signed-off-by: Ludovic Courtès <ludo@gnu.org>
For the GDBM and Marshal dictionary adapters to be useful with their
default settings, we must initialize their data files during the package
build: upstream would initialize them lazily, but that doesn't work with
an immutable installation directory (at least, not without more complex
patches). Otherwise, we would always end up rebuilding the dictionary at
startup, which is “slow” and “not recommended”.
* gnu/packages/patches/ruby-anystyle-fix-dictionary-populate.patch: New
patch.
* gnu/local.mk (dist_patch_DATA): Add it.
* gnu/packages/ruby.scm (ruby-anystyle)[patches]: Use it.
[arguments]<#:phases>: Add 'populate-dictionaries' phase.
Signed-off-by: Ludovic Courtès <ludo@gnu.org>
* gnu/packages/patches/ruby-anystyle-data-immutable-install.patch: New
patch.
* gnu/local.mk (dist_patch_DATA): Add it.
* gnu/packages/ruby.scm (ruby-anystyle-data)[patches]: Use it.
Signed-off-by: Ludovic Courtès <ludo@gnu.org>
