| Commit message (Collapse) | Author | Age | Files | Lines |
|
|
|
|
|
|
| |
Since rm_rf_at now removes the object being pointed at, rm_rf doesn't
have to try doing that again.
Signed-off-by: Fabian Groffen <grobian@gentoo.org>
|
|
|
|
| |
Signed-off-by: Fabian Groffen <grobian@gentoo.org>
|
|
|
|
|
|
|
|
|
| |
Because FEATURES is an ISTR (incremental string) when a default it set,
we're never able to loose the default, so don't set it. This causes a
behavioural change of "nodoc noman noinfo" no longer being (always) set.
qmerge in particular will hence install /usr/share/{doc,man,info} again.
Signed-off-by: Fabian Groffen <grobian@gentoo.org>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Reworked qmerge install_mask handling to support negative entries in
INSTALL_MASK, which "undo" (part of) a previous masking.
This should implement the same behaviour as
https://archives.gentoo.org/gentoo-portage-dev/message/29e128a9f41122fa0420c1140f7b7f94
Along the way some behavioural changes were made:
- pretend mode now prints what would be merged (use with -y to be
useful)
- directory objects are recorded and merged again
Bug: https://bugs.gentoo.org/656318
Signed-off-by: Fabian Groffen <grobian@gentoo.org>
|
|
|
|
|
|
|
|
| |
To better emulate 'rm -Rf' behaviour, if path points to a file, just
remove it. If it points to a directory, remove that directory instead
of leaving the (now) empty directory behind.
Signed-off-by: Fabian Groffen <grobian@gentoo.org>
|
|
|
|
|
|
|
|
| |
When --name-only is in effect, don't print the space separating the
package and its description, for we won't print the latter.
Bug: https://bugs.gentoo.org/672764
Signed-off-by: Fabian Groffen <grobian@gentoo.org>
|
|
|
|
|
|
|
| |
Continuation of bug #668418, atom_compare is strict about SLOT matches.
Bug: https://bugs.gentoo.org/668418
Signed-off-by: Fabian Groffen <grobian@gentoo.org>
|
|
|
|
|
|
|
|
| |
atoms in dependency strings can contain built-with-use declarations in
bracket notation, so ensure we ignore that when feeding this into
atom_explode.
Bug: https://bugs.gentoo.org/668418
|
| |
|
|\
| |
| | |
qmerge: support for more decompresssion programs
|
|/
|
|
|
|
|
| |
Support additional compression programs like zstd or lz4 for BINPKG_COMPRESS.
Also make the gzip test less http://www.onicos.com/staff/iz/formats/gzip.html
and the bzip2 test more https://en.wikipedia.org/wiki/Bzip2#File_format
restrictive.
|
|
|
|
|
|
|
|
| |
Portage can create binpkgs compressed differently than with bzip2, so
try to probe the file quickly before untarring, so we can use the
correct flag with tar.
Bug: https://bugs.gentoo.org/660508
|
|
|
|
|
|
|
|
|
| |
Check if the python atom check is going to work, if not, fall back to
full pre-cooked files. If we just check for a portage tree in the
filesystem, it may be that we end up without python support for portage,
which would then obviously mismatch.
Bug: https://bugs.gentoo.org/661762
|
|
|
|
| |
Signed-off-by: Fabian Groffen <grobian@gentoo.org>
|
|\
| |
| | |
Fix code snippets highlighting in README
|
|/ |
|
|
|
|
|
| |
Make this test run on platforms with a different portdir, or without
Portage (such as Travis). This should avoid regressions like bug #567336
|
|
|
|
|
|
|
|
| |
We need to keep on searching until we reached the end of the
package/version string, since package names may contain things which are
valid versions, as long as they don't end with them.
Bug: https://bugs.gentoo.org/567336
|
| |
|
| |
|
| |
|
| |
|
| |
|
| |
|
| |
|
| |
|
| |
|
|
|
|
| |
Bug: https://bugs.gentoo.org/619988
|
| |
|
| |
|
| |
|
|
|
|
|
| |
This reverts the accidential commit to Makefile from
054b993a9433a45c00086fb2841a8150d85eb437.
|
| |
|
|
|
|
|
|
|
| |
A bit overenthusiastic check for files in the current directory caused
weird directory prefixes to be ignored.
Bug: https://bugs.gentoo.org/652720
|
|
|
|
| |
Bug: https://bugs.gentoo.org/652312
|
|
|
|
| |
While at it document the date formats in the man-page.
|
|
|
|
|
|
|
|
| |
It's not enough to find a hyphen followed by something that looks like a
version, it needs to be verified it really is, else it is part of the
package name.
Bug: https://bugs.gentoo.org/653032
|
| |
|
| |
|
|
|
|
|
|
|
|
|
|
|
| |
This commit changed the logic from fetching Packages when -f is set, to
forcing when -F is set. This is contradicting the documented behaviour
and broke the tests.
It is not clear to me anymore why this change was necessary, so I'm
reverging it.
This reverts commit eedb0ea266b78efd84bc34bc2c6602fb713ba590.
|
|
|
|
|
|
| |
When using optimisation, the compiler does something to a char pointer
it doesn't do to a void pointer, so use a couple of casts to avoid a
clobber warning.
|
| |
|
|
|
|
|
|
| |
The static tests fell off the radar because actual differences did not
result in failure (e.g. Travis being green and all), also the sed magic
was too loose, causing a wrong input file to be computed
|
|
|
|
|
| |
atom_explode/basic test wasn't run due to no dependencies in Makefile,
hence I didn't see this before
|
|
|
|
|
| |
as side-effect of just including everything (non-library style) the
compiler complains when static functions aren't used, tell it not to
|
| |
|
|
|
|
|
|
|
|
| |
Extract the version part of the atom, such that we can more reliably
parse the version components without worrying about accidentially taking
parts of the package name.
Bug: https://bugs.gentoo.org/526596
|
|
|
|
|
|
|
|
| |
If we start counting parallel merges only once we found something
interesting, we might not be aware of other emerges at that time,
drawing wrong conclusions when seeing a termination message.
Bug: https://bugs.gentoo.org/603024
|
|
|
|
|
|
|
|
| |
For consistency with almost all other output, use sorted output variant
for qfile as well. The little sorting/memory overhead due to this is
probably ok, given how much it is used already.
Bug: https://bugs.gentoo.org/607498
|
| |
|