As regards the (commented) URIs referring to the available patchlevels:
- employ https:// rather than ftp://
- ensure that the URI request path consistently matches the version
- remove the URI altogether for snapshot ebuilds
Signed-off-by: Kerin Millar <kfm@plushkava.net>
Signed-off-by: Sam James <sam@gentoo.org>
Drop 5.2_p37, along with the files that it was uniquely referencing.
Signed-off-by: Kerin Millar <kfm@plushkava.net>
Signed-off-by: Sam James <sam@gentoo.org>
This commit introduces 5.2_p37-r4 and bash-5.3_p3-r1, as well as
revising bash-9999. The effective changes are described herewith.
Install a new /etc/profile.d drop-in, named "00-prompt-command.sh". It
serves to initialise the PROMPT_COMMAND variable as early as is
conveniently possible, provided that bash is acting as a login shell.
Install a revision of the /etc/bash/bashrc file that refrains from
initialising the PROMPT_COMMAND variable, provided that bash is acting
as a login shell. Along with the aforementioned change, this amounts to
a simplified interpretation of GitHub PR #42626. Though not a common
practice, some /etc/profile.d drop-ins (such as vte-2.91.sh) need to be
able to manipulate PROMPT_COMMAND and had hitherto been stymied by
bashrc proceeding to unconditionally clobber it.
Finally, refrain from installing 15-gentoo-bashrc-check.bash to
/etc/bash/bashrc.d. At the time of its introduction, I recommended that
it should exist for no longer than a year. Its removal is overdue.
Reported-by: Krzesimir Nowak <knowak@microsoft.com>
Part-of: https://github.com/gentoo/gentoo/pull/42626
Closes: https://github.com/gentoo/gentoo/pull/42626
Signed-off-by: Kerin Millar <kfm@plushkava.net>
Signed-off-by: Sam James <sam@gentoo.org>
The top of FILESDIR has become a little unwieldy. In advance of
submitting a patch that will introduce yet more files to be tracked, I
thought that it would be a good opportunity to conduct a minor
re-organisation. To that end, this commit relocates the skeleton files
to a sub-directory.
Signed-off-by: Kerin Millar <kfm@plushkava.net>
Signed-off-by: Sam James <sam@gentoo.org>
Bug 944459 was fixed together with bug 925558 in commit a4bfadcafd
("app-shells/rc: fix build failure, port to C23 For libedit, we need to
carefully truncate size_t to int but that's number of lines, so it
should work in sane situations").
Bug: https://bugs.gentoo.org/925558
Closes: https://bugs.gentoo.org/944459
Signed-off-by: Petr Vaněk <arkamar@gentoo.org>
Using system sqlite turns out to have a significant performance
impact. Until the actual cause is found, use the bundled sqlite per
default.
Also remove the empty CRATES definition.
Bug: https://bugs.gentoo.org/959120
Signed-off-by: Florian Schmaus <flow@gentoo.org>
The package used greadme.eclass before, but was, for some unknown reason,
switched back to readme.gentoo-r1.eclass with
5cc79e4bfc ("app-shells/atuin: add 18.6.1"). However, this broke
multiple things:
- readme.gentoo-r1.eclass is not as flexible with optional README files
as greadme.eclass, causing https://bugs.gentoo.org/956351 which was fixed with
d1911f241f ("app-shells/atuin: fix call to readme.gentoo_create_doc
IUSE="client"")
- the README text was copied verbatim from the bash here-doc into a
file, resulting in wrong quotes being shown
- the readme text is only sensible if USE=client and hence was only
conditionally shown if this the 'client' USE flag was enabled
Ultimately, greadme.eclass is more sophisticated as
readme.gentoo-r1.eclass and can be used in every situation where
readme.gentoo-r1.eclass is used.
Signed-off-by: Florian Schmaus <flow@gentoo.org>
