IF YOU WOULD LIKE TO GET AN ACCOUNT, please write an
email to Administrator. User accounts are meant only to access repo
and report issues and/or generate pull requests.
This is a purpose-specific Git hosting for
BaseALT
projects. Thank you for your understanding!
Только зарегистрированные пользователи имеют доступ к сервису!
Для получения аккаунта, обратитесь к администратору.
Directory size is an install-time filesystem specific implementation
detail, build-time just doesn't have a clue. If we want to try taking
directory sizes into account, this needs to happen in the disk space
checking at install time.
It also helps making builds a bit more reproducable (PR #229)
(cherry picked from commit 2cf7096ba534b065feb038306c792784458ac9c7)
The canonicalization is based on the code lifted from tools/debugedit.c
The new behaviour is enabled by default and controlled by
%_canonicalize_symlinks macro.
While supported of course, absolute symlinks are nasty in many ways
so issue a warning to packagers. Update two affected test-cases.
Related to #668.
(cherry picked from rpm.org commit 8359ded6784579b9f2527a09888132034783e1de)
when the files exist during build time.
The hash will never be used for verification anyway.
This helps making packages build more reproducibly.
(cherry picked from rpm.org commit 8cda485e1150dc6be990a7aa4496da848154cf55)
It might be a rare condition but it /can/ happen, and previously that would cause '\0' written out of bounds (at negative array offset). Also leave room for the terminating '\0' at the end of buffer when calling readlink(), previously a link exactly the size of buffer would've been silently truncated.
(cherry picked from rpm.org commit c499a0b912629eeeabc3554d3e3a95c088e96422)
Spec files have a lot of metadata about a project. However one of the
most key components is the upstream version control system which was
notably lacking.
Resolve this by adding a "VCS" key. There is no specification
for contents of this key, given that the set of version control
systems (and features thereof) are not well-defined. However,
recommendations are:
* git: This URL should be in a form that can be passed to "git clone",
with the additional feature that an optional fragment identifier "#foo"
denotes a branch or tag.
These are the cases where even the release was not needed (so, the
disttag is not needed, too), or one case where the filename is
contructed (and it doesn't include the disttag).
Now grep -Ee 'headerNVR[^D]' will show the remaining non-trivial
cases, where adapting to disttags may be needed.
Not touched:
Requires: x < E:V-R[:D]
Obsoletes: x > E:V-R[:D]
Requires: x > E:V-R:D (with a disttag)
There are actually no such dependencies in Sisyphus now.
With this change, there are not so many failing tests from
rpminstall-tests-1.0-alt4 (or equivalently from the predisttag branch
there) when run in a system with an "old" rpm and the "new" rpm-build
$ rpm -q rpm --lastchange
* Пт окт 05 2018 Gleb F-Malinovskiy <glebfm@> 4.13.0.1-alt4
- Add _allow_deps_with_beginning_dot macro to allow dependencies
beginning with a dot character in spec file (vseleznv@).
$ ./makeme.sh -j4 SHORT=1
...
XFailed tests:
installable_dummyDisttag_with_reqEqualEpoch (external/strict-old-old dep & future pkg)
noninstallable_dummyDisttag_with_conflEqualEpoch (external/strict-old-old dep & future pkg)
installable_virtDummyDisttag_with_reqEqualEpoch (external/strict-old-old dep & future pkg)
noninstallable_virtDummyDisttag_with_conflEqualEpoch (external/strict-old-old dep & future pkg)
installable_dummyDisttag_with_reqEqual (elusive external/strict-old-old dep & future pkg)
noninstallable_dummyDisttag_with_conflEqual (elusive external/strict-old-old dep & future pkg)
installable_virtDummyDisttag_with_reqEqual (elusive external/strict-old-old dep & future pkg)
noninstallable_virtDummyDisttag_with_conflEqual (elusive external/strict-old-old dep & future pkg)
obsoleted_dummyDisttag_with_obsolEqualEpochDisttag (unrealistic obsoleting disttag)
obsoleted_dummyDisttag_with_obsolEqualDisttag (unrealistic obsoleting disttag)
installable_provVR3Disttag_with_reqEq_VR3 (elusive external/strict-old-old dep & future pkg)
Failed tests:
noninstallable_provOnlyV_with_reqVR
noninstallable_provOnlyV_before_reqVR
installable_provOnlyV_with_conflVR
installable_provOnlyV_before_conflVR
Except for the underspecified Provides problem, the insteresting cases
are the cases of a = dependency.
We can think of a Conflicts: subpkg = E:V-R, which wouldn't work,
but many such cases (if there are any) must be mitigated by the change
of the value of %%EVR to include the disttag (in another commit).
So, the "old" and "new" rpm behave much more similar w.r.t. to the
packages built by the new rpm-build. (This is what we wanted.)
- Implemented limited support for large files: a 2GB+ file can now be packaged,
but the total size of uncompressed cpio payload is capped at 4 GB.
- Automatically downgrade LZMA compression levels 7-9 -> 6 on small payloads.
This is probably the last change of such kind. There are many other
instances left, and fixing them all is hopeless. On the other hand,
the way mod 2^32 arithmetic works, although technically not always
well defined, is to our advantage. I suggest that only user-visible
discrepancies further be fixed.
The following comparison to st_size looks particularly bad.
The code turns out to be unused!
lib/signature.c:
> verifySizeSignature(const char * datafile, int_32 size, /*@out@*/ char * result)
> [...]
> if (size != st.st_size) {
> sprintf(result, "Header+Archive size mismatch.\n"
This turns out to be a bit easier than I thought yesterday. When mmap
fails in lib/fsm.c, it falls back to file descriptor IO. So by simply
enabling AC_SYS_LARGEFILE, I'm already able to package a 3G file, which
can be written like this:
perl -e 'seek STDOUT,3<<30,0 and print 1 or die' >out
The nearest goal is to be able to package files up to 4G, on both 32-bit
64-bit architectures, provided that cpio payload size also does not
exceed 4G. More specifically, the limit is 2^32-2, while 2^32-1 remains
special (this is not my own invention, I adopt it from rpm.org).
If the limit is exceeded, rpmbuild should fail gracefully instead of
writing malformed packages with truncated integer tags.
(Supporting 4G+ files is a distant goal - neither rpm-4.0 nor apt-rpm
can handle them yet. Besides, it's not clear if producing such big
packages is even a good idea, unless one wishes to package p0rn.)
If RPM_STRICT_INTERDEPS environment variable is set and not empty, this
function replaces every strict requirement on NEVR with
".${RPM_STRICT_INTERDEPS}-NEVR", and adds to every package which
provides NEVR ".${RPM_STRICT_INTERDEPS}-NEVR" provide if some other
package requires this ".${RPM_STRICT_INTERDEPS}-NEVR".
This is needed to able to build a source package to different repo
branches with the same NEVR, and to rebuild the source package as well
with same NEVR within one branch and avoid interdep collisions.
SIGPIPE SIG_IGN handler was installed before the fork, which means that,
in autodep scripts, SIGPIPE was ignored as well. This is why in
commands like
cmd1 | cmd2
cmd1 was not killed graceully with SIGPIPE, but instead writing to cmd2
resulted in EPIPE. For which some commands apparently were not ready.
This fixes messages like
/usr/lib/rpm/files.req: line 33: echo: write error: Broken pipe
Note that checkHardLinks function is now removed. It was unclear
whether it was supposed to verify %lang attributes (returning non-zero
on error) or indicate if all hardlinks are packaged within the package.
It turns out that only a single package in our repo has
PartialHardlinkSets dependency:
$ cd /ALT/Sisyphus/files/x86_64/RPMS/
$ rpm -qp --qf '[%{NAME}\t%{REQUIRENAME}\n]' *.rpm |fgrep 'PartialHardlinkSets'
$ cd /ALT/Sisyphus/files/noarch/RPMS/
$ rpm -qp --qf '[%{NAME}\t%{REQUIRENAME}\n]' *.rpm |fgrep 'PartialHardlinkSets'
freeciv-common rpmlib(PartialHardlinkSets)
$
This probably means that freeciv-common has hardlinks with different
%lang attributes (which probably was supposed to be an error). So
the whole issue should be reconsidered. A leave XXX marks in the
code and suggest new PartialHardlinkSets implementation (however,
the dependency is not being added yet).
When addFile() is called with file path obtained from specfile,
various checks must be performed, so that only valid paths can
enter the package. However, when a path is passed down to fts,
it is no longer possible to obtain ill-formed paths. So a lot
of lstat calls for intermediate path components can be saved.
Since the list of files is sorted, it is possible to avoid quadratic
behaviour by implementing merge-like intersection.
However note that directory names are not sorted (this is why it is
needed to build full file name for strcmp).
Tested with gnome-icon-themes-oxygen-refit2-2.3-alt1.src.rpm,
3 subpackages each 13-16K files. Old behaviour: 18 seconds,
new behaviour: less than 1 second.
Some code (e.g. apt/genpkglist) explicitly relies on the fact that
header file list is represented with baseNames+dirNames+dirIndexes
arrays. Thus, generating legacy headers might have issues, and should
be disabled.
- implemented post-transaction filetriggers, loosely based on filetriggers.patch
from Mandriva Linux (see %_rpmlibdir/posttrans-filetriggers for details)
- implemented %_rpmlibdir/0ldconfig.filetrigger, so that packages with
shared libraries need not to invoke ldconfig(1) in they %%post-scriptlets
