Jan Rękorajski [Tue, 12 Mar 2013 17:28:47 +0000 (18:28 +0100)]
- fix for bug LP#1104474
- always use inode hash as generated in build/files.c:genCpioListAndHeader()
to prevent overflow of 64bit inode numbers
- rel 43
Jan Rękorajski [Sun, 2 Dec 2012 13:58:56 +0000 (14:58 +0100)]
- removed namespace check in rpmal.c, it was unneccessary as the
comparison was on unparsed names and broke name-to-provides check
during upgrades
- rel .5
Jan Rękorajski [Fri, 30 Nov 2012 20:51:07 +0000 (21:51 +0100)]
- fix the root cause behind infinite loop during ntpd install
On Fri, 30 Nov 2012, Jeffrey Johnson wrote:
[...]
> Yes: the iterator loop index on dependency sets is global.
>
> Which means that if something decrements the iterator index
> lower down, then the high level will continuously repeat.
And this exactly happened in lib/rpmal.c:rpmalAllSatisfiesDepend()
for 'provides' set because ntpd provided 'ntp' twice in different
manespaces.
Jan Rękorajski [Fri, 30 Nov 2012 12:26:03 +0000 (13:26 +0100)]
- check namespaces of dependencies in _rpmtsCheck, don't compare strings blindly
This patch fixes a bug with ntpd package we encoutered:
- ntpdate has "Conflicts: ntp < 4.2.0-3"
- ntpd has "Provides: ntp = 4.2.4" and "Provides: user(ntp)"
now, if ntpdate is installed then attempt to install ntpd causes
_rpmtsCheck to compare "C: ntp" to both "P: ntp" AND THEN "P: user(ntp)"
due to lack of dependency namespace check. Side effect of this is
infinite loop in _rpmtsCheck due to inner workings of rpm dependency
iterators.
Jan Rękorajski [Sun, 11 Nov 2012 10:10:26 +0000 (11:10 +0100)]
- add deps for building xz compressed archives to rpm-build package
- fully replaces default macrofiles list to have proper order of loading
them: libdir -> sysconfigdir -> homedir
- rel 35
Jan Rękorajski [Thu, 25 Oct 2012 22:43:36 +0000 (00:43 +0200)]
- proper fix to mdvbz62979 is not copying description and summary to subpackages
Copying description to subpackages on they creation time caused
description parser to remove all but the last description translation,
even removed LANG=C default descriptions from subpackages.
Jan Rękorajski [Thu, 25 Oct 2012 09:51:57 +0000 (11:51 +0200)]
- moved __nofsync definition
Rationale:
With %{load:...} being interpreted immediately -- not lazily --
within a macro init file, a failure to load causes immediate
loading termination silently.
Jan Rękorajski [Mon, 22 Oct 2012 13:13:23 +0000 (15:13 +0200)]
- disable hmac when veryfying file digests
- rel 24
Disable hmac during digest calculation, since rpm package files
contain plain md5sums, hmac support is useless, see:
http://lists.pld-linux.org/mailman/pipermail/pld-devel-en/2012-October/023193.html
Making all in tests
warning: edos-test-1-0.src.rpm: Header V4 DSA signature: NOKEY, key ID 2039b291
*** glibc detected *** /home/users/builder/rpm/BUILD/rpm-5.4.10/.libs/lt-rpm: free(): invalid size: 0x00000000025df940 ***
*** glibc detected *** /home/users/builder/rpm/BUILD/rpm-5.4.10/.libs/lt-rpm: corrupted double-linked list: 0x00000000025aeca0 ***
If rpm was built without neon, then during build, on PLD builders,
the error shown above manifested itself on random test packages
and then just-built rpm got stuck at a futex.
Jan Rękorajski [Wed, 3 Oct 2012 13:03:15 +0000 (15:03 +0200)]
- build with internal lua and enable extra lua libs, fixes
http://lists.pld-linux.org/mailman/pipermail/pld-devel-en/2012-September/023154.html
- rel 17
Elan Ruusamäe [Sat, 29 Sep 2012 18:42:39 +0000 (21:42 +0300)]
use db package name in deps
otherwise we end up in funny situation that rpm reqires db > 4.7, but
installs db 5.3 to satisfy it:
Processing dependencies...
db4.7-4.7.25.4-7.i686 obsoleted by db4.7-4.7.25.4-8.i686
db4.7-utils-4.7.25.4-7.i686 obsoleted by db4.7-utils-4.7.25.4-8.i686
orphaned rpm-lib-4.5-69.i686 marks db5.3-5.3.21.0-1.i686 (cap db >= 4.7.25)