Go and other cross-compilation toolchains ship binaries for architectures
not previously listed (LOONGARCH, S390, ALPHA, SH64, SH). scanelf reports
these verbatim; add them so metadata extraction does not fail on packages
like dev-lang/go that include cross-arch ELF files.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
scanelf reports UNKNOWN_TYPE for non-standard or non-ELF binaries (e.g.
proprietary Teensy tools in arduino-bin). elf_class_strtoint() raised
ValueError which aborted metadata extraction for the whole package.
Skip unrecognised ELF class entries instead.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
When rescue spmsync truncates the temp file to zero bytes before writing
XPAK metadata, calling get_binpkg_format() triggers seek(-16, 2) on an
empty file which prints an [Errno 22] stderr warning for every package.
Skip the format detection when the file is empty — it can only be xpak.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Two fixes for `equo rescue spmsync` on a system without an existing
installed packages DB:
1. client/interfaces/methods.py: Stop falling back to a GenericRepository
(in-memory, read-only for handlePackage) when the DB file is absent.
Instead let SQLite create the file; call initializeRepository() on a
freshly created DB so validate() succeeds. Corrupted DBs still fall
back to RAM via the existing DatabaseError/SystemDatabaseError path.
2. spm/portage_plugin: Also catch OSError (not just tarfile.ReadError)
when uncompress_tarball fails on a pure-XPAK temp file; Python 3.14
raises OSError in this case rather than tarfile.ReadError.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Python 3.14 rejects action='store_true' on positional arguments.
The add_argument calls in _show_help() are display-only (never parsed),
so the action kwarg was unnecessary.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
- sql.py: convert setCategoryDescription() INSERT loop to executemany()
(only remaining INSERT-in-loop; rest of sql.py already used executemany)
- spm/plugins/skel.py: add from __future__ import annotations and type
annotations to all 25 abstract interface methods (return types, param types)
using Python 3.10+ union syntax (X | None instead of Optional[X])
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Simplify const_isfileobj(), const_isnumber(), const_debug_write()
and remove unreachable code in rigo/rigo/utils.py::prepare_markup().
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
- Fix counter_tick(): portage 3.x removed the root/mycpv arguments,
now a zero-argument call returning the new counter value directly
- Add IDEPEND (EAPI 8 install-time deps) to xpak_entries, extraction
dict, _calculate_dependencies and dep_keys (mapped to bdepend_id)
- Add GPKG binary package format support (.gpkg.tar) alongside xpak:
extract_package_metadata() detects format via get_binpkg_format() and
uses PortageGPKG.unpack_metadata()/decompress() for gpkg packages;
append_metadata_to_package() uses PortageGPKG.update_metadata() for gpkg
- Add "gpkg.tar" to binary_packages_extensions()
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
- Replace deprecated Thread.isAlive() with is_alive() (Python 3.12)
- Fix invalid escape sequences in regex strings (SyntaxWarning → SyntaxError)
- Replace removed unittest aliases: assertEquals→assertEqual,
assertNotEquals→assertNotEqual, assert_→assertTrue,
assertRaisesRegexp→assertRaisesRegex (Python 3.12)
- Replace portage's removed clear_caches() with flush_cache()
- Fix locks.py: reset TLS state on exception in _file_lock_create
to prevent stale shared-lock state poisoning subsequent acquisitions
- Skip os.chown() when gid is None or caller lacks permission
- Fix FastRSS attribute ordering for consistent minidom re-parse behavior
- Update test expected XML to match Python 3.8+ minidom attribute order
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
FakeOutFile and LogFile need .buffer like Python 3 "text" file objects
which is needed when in rigo standard output/error is replaced, and then
it reaches Portage which does this:
if sys.hexversion >= 0x3000000 and fd in (sys.stdout, sys.stderr):
fd = fd.buffer
fd.write(mystr)
(/usr/lib64/python3.6/site-packages/portage/util/__init__.py).
Entropy internal code did not need this.
Note, after this commit, changes done previously:
1)
commit 0869912ec4
[entropy.spm] Rigo related Python 3 fix
File "/usr/lib64/python3.6/site-packages/entropy/spm/plugins/interfaces/portage_plugin/__init__.py", line 101, in _pusher
self._std.buffer.write(chunk)
AttributeError: 'FakeOutFile' object has no attribute 'buffer'
2)
commit 8700aade27
[entropy.spm, rigo] Rigo related Python 3 fixes
1)
File "/usr/lib/python-exec/python3.6/RigoDaemon_app.py", line 362, in _pusher
fobj.write(chunk)
TypeError: write() argument must be str, not bytes
2)
File "/usr/lib64/python3.6/site-packages/entropy/spm/plugins/interfaces/portage_plugin/__init__.py", line 77, in __init__
self.buffer = Writer(self, self._std.buffer)
AttributeError: 'FakeOutFile' object has no attribute 'buffer'
(...)
could be likely reverted; not tested, it could be better to have them anyway to
avoid futher re/encoding/checks if conversions are needed (subjective).
Fixes bug 5899.
File "/usr/lib64/python3.6/site-packages/entropy/spm/plugins/interfaces/portage_plugin/__init__.py", line 101, in _pusher
self._std.buffer.write(chunk)
AttributeError: 'FakeOutFile' object has no attribute 'buffer'
1)
File "/usr/lib/python-exec/python3.6/RigoDaemon_app.py", line 362, in _pusher
fobj.write(chunk)
TypeError: write() argument must be str, not bytes
2)
File "/usr/lib64/python3.6/site-packages/entropy/spm/plugins/interfaces/portage_plugin/__init__.py", line 77, in __init__
self.buffer = Writer(self, self._std.buffer)
AttributeError: 'FakeOutFile' object has no attribute 'buffer'
With Python 3, data in sqlite from packagesignatures.gpg column was binary on
read (because it was binary on write) but a string was actually required.
The problem was visible with GPG signed packages: when injected into database
using eit with Python 3, it would then explode during equo update/install on
client side.
error from Python 2:
File "/usr/lib64/python2.7/site-packages/entropy/client/interfaces/package/actions/fetch.py", line 1063, in do_signatures_validation
if hash_val in signatures:
TypeError: writable buffers are not hashable
error from Python 3:
File "/usr/lib64/python3.6/site-packages/entropy/client/interfaces/package/actions/fetch.py", line 1023, in do_compare_gpg
tmp_f.write(hash_val)
TypeError: write() argument must be str, not bytes
Added a new way to retrieve list of kernel available
that uses directory /etc/kernels/availables/{sabayon,sabayon-lts}.
If this directory is present is not used current logic for
retrieve the list of kernel.
Under /etc/kernels/avaiables/ directory files with 'lts' extension
are used for retrieve list of LTS kernels.
An user could return his kernel list just if he creates and
maintains his file in a similar format:
$# cat /etc/kernel/availables/mykernel
sys-kernel/linux-sabayon:4.9
sys-kernel/linux-mykernel:4.4
This logic now is used as an additional feature after retrieve
the existing kernels based on virtual/linux-binary or
virtual/linux-binary-lts.
ConfigurationFiles._load_maybe_add (equo.client.misc): path is in bytes here
_std_write (entropy.output): fixes crash in equo conf update with "Replace
original with update" when there is an error in which case msg can be bytes:
File ".../entropy/lib/entropy/tools.py", line 687, in movefile
print_generic("!!! Failed to move", src, "to", dest)
File ".../entropy/lib/entropy/output.py", line 596, in print_generic
_std_write(msg, stderr = stderr)
File ".../entropy/lib/entropy/output.py", line 493, in _std_write
obj.write(msg)
Function callers, in order (during eit commit):
1. filename='lib/entropy/server/interfaces/main.py', function='switch_default_repository', code_context=[' self.close_repositories()\n'], index=0)
2. filename='server/eit/commands/command.py', function='_call_exclusive', code_context=[' server.close_repositories()\n'], index=0)
3. filename='lib/entropy/client/interfaces/client.py', function='destroy', code_context=[' self.close_repositories(mask_clear = False)\n'], index=0)
4. filename='lib/entropy/server/interfaces/main.py', function='destroy', code_context=[' self.close_repositories()\n'], index=0)
Third one triggers packages set synchronization (which would cause "dictionary
changed size during iteration"), and marks sets as being synchronized.
Fourth does not trigger sets synchronization, so closes all repositories
without new ones being opened in the meantime.
Side note:
It's similar to this in "client," lib/entropy/client/interfaces/methods.py:
def close_repositories(self, mask_clear = True):
...
# list() -> python3 support
for item, val in list(repo_cache.items()):
...
repo_cache.pop(item).close(_token = repository_id)
...
but (based on shallow look), it doesn't do as much magic; just calls .pop()
(not sure if there are similar side effects in close() there, though), so the
explanation may possibly apply to the lib/entropy/server function only.
Mario Fetka
Daniele Rondina
Joost Ruis
Gokturk Yuksek
Sławomir Nizio
Ettore Di Giacinto
Geaaru
Ettore Di Giacinto