From qboosh at pld-linux.org  Thu Nov  1 14:10:51 2012
From: qboosh at pld-linux.org (Jakub Bogusz)
Date: Thu, 1 Nov 2012 14:10:51 +0100
Subject: [packages/filesystem] - can't use any non-root groups in this
 package and not depend on setup package, do similar workardo
In-Reply-To: <aefd25979c44a66fc4eecfc32a27d2d119f31d76_refs_heads_master@pld-linux.org>
References: <84956be926dc0639d1c8c3c0092fab46e402d86e_refs_heads_master@pld-linux.org>
 <aefd25979c44a66fc4eecfc32a27d2d119f31d76_refs_heads_master@pld-linux.org>
Message-ID: <20121101131051.GA25496@mail>

On Wed, Oct 31, 2012 at 07:42:21AM +0100, glen wrote:
> commit aefd25979c44a66fc4eecfc32a27d2d119f31d76
> Author: Elan Ruusam?e <glen at delfi.ee>
> Date:   Wed Oct 31 08:41:13 2012 +0200
> 
>     - can't use any non-root groups in this package and not depend on setup package, do similar workardound as FHS package

And what is the problem with requiring setup here?
setup requires just FHS, not filesystem.

Your change causes rpm -V to report /var/log/archive group inconsistency on every system.


> diff --git a/filesystem.spec b/filesystem.spec
> index 88b1eec..c2f06c0 100644
> --- a/filesystem.spec
> +++ b/filesystem.spec
> @@ -1,3 +1,6 @@
> +# NOTE
> +# - do not use any other user/group than "root", as then we have to depend on "setup" package.
> +#   see the gid_xxx macros and post scriptlet
>  
>  # disable rpm generated debug package, we handle it differently here
>  %define		_enable_debug_packages	0
> @@ -8,7 +11,7 @@ Summary:	Common directories
>  Summary(pl.UTF-8):	Wsp?lne katalogi
>  Name:		filesystem
>  Version:	4.0
> -Release:	18
> +Release:	19
>  License:	GPL
>  Group:		Base
>  BuildRequires:	automake
> @@ -39,6 +42,9 @@ BuildRoot:	%{tmpdir}/%{name}-%{version}-root-%(id -u -n)
>  # directory for *.idl files (for CORBA implementations)
>  %define		_idldir		/usr/share/idl
>  
> +# we have to use numeric uids/groups. see comment beginning of the spec
> +%define		gid_logs	124
> +
>  %description
>  This package contains common directories for packages that extend some
>  programs functionality, but don't require them themselves.
> @@ -151,6 +157,9 @@ if posix.stat("/usr/include/X11", "type") == "link" then
>  end
>  %endif
>  
> +%post -p <lua>
> +posix.chown("/var/log/archive", 0, %{gid_logs})
> +
>  %files -f %{name}.lang
>  %defattr(644,root,root,755)
>  %dir /boot/efi
> @@ -223,7 +232,7 @@ end
>  %dir /var/lib/color
>  %dir /var/lib/color/icc
>  %attr(700,root,root) %dir /var/lock/subsys
> -%attr(751,root,logs) %dir /var/log/archive
> +%attr(751,root,root) %dir /var/log/archive
>  %dir %{_aclocaldir}
>  %dir %{_desktopdir}
>  %dir %{_desktopdir}/docklets
> ================================================================
> 
> ---- gitweb:
> 
> http://git.pld-linux.org/gitweb.cgi/packages/filesystem.git/commitdiff/bca6232882d14ac3d685f02dfdaacfa6295ec948
> 
> _______________________________________________
> pld-cvs-commit mailing list
> pld-cvs-commit at lists.pld-linux.org
> http://lists.pld-linux.org/mailman/listinfo/pld-cvs-commit

-- 
Jakub Bogusz    http://qboosh.pl/


From qboosh at pld-linux.org  Thu Nov  1 14:13:22 2012
From: qboosh at pld-linux.org (Jakub Bogusz)
Date: Thu, 1 Nov 2012 14:13:22 +0100
Subject: systemd unit options [[packages/corosync] make
 sysconfig/corosync-notifyd usable for systemd]
In-Reply-To: <20121030074023.GA1826@home.lan>
References: <9c0f7a6e7d1349446f09273cc3272f77863bcf7c_refs_heads_master@pld-linux.org>
 <ac647687b5d8c7bfb01cd7fc24ef36c23b10cc23_refs_heads_master@pld-linux.org>
 <20121029163518.GA9794@mail> <20121030074023.GA1826@home.lan>
Message-ID: <20121101131322.GB25496@mail>

On Tue, Oct 30, 2012 at 08:40:23AM +0100, Jan R?korajski wrote:
> On Mon, 29 Oct 2012, Jakub Bogusz wrote:
> 
> > On Mon, Oct 29, 2012 at 04:23:56PM +0100, jajcus wrote:
> > > commit ac647687b5d8c7bfb01cd7fc24ef36c23b10cc23
> > > Author: Jacek Konieczny <j.konieczny at eggsoft.pl>
> > > Date:   Mon Oct 29 16:22:10 2012 +0100
> > > 
> > >     make sysconfig/corosync-notifyd usable for systemd
> > >     
> > >     The sysconfig file used by our init script would not be usable for
> > >     the sytemd unit got upstream.
> > 
> > > --- a/corosync-notifyd.sysconfig
> > > +++ b/corosync-notifyd.sysconfig
> > > @@ -1,12 +1,3 @@
> > >  
> > > -# Send SNMP traps on all events.
> > > -SEND_SNMP_TRAPS=no
> > > -
> > > -# SNMP Manager IP address (defaults to localhost).
> > > -#SNMP_MANAGER=
> > > -
> > > -# Send DBUS signals on all events.
> > > -SEND_DBUS_SIGNALS=yes
> > > -
> > > -# Log all events.
> > > -LOG_ALL_EVENTS=no
> > > +# log and signal over DBus all events
> > > +OPTIONS=-l -d
> > 
> > So systemd is so great everyone must learn all daemon options again,
> > like it was before SysV scripts/sysconfig were invented?
> > 
> > There are no upstream solutions?
> 
> Only simple environment variables usage.
> IMO it's not that big of a problem as often our sysconfig definitions
> contain only a random subset of daemon options forcing to mess with
> init script. With systemd it's at least clear - one have to read the
> manual if (s)he wants some special options.
> 
> What we can do in this case is to document the most usefull options
> in the sysconfig files.

That would be acceptable - document at least those options, which were
supported by SysV init script and add a pointer to documentation
(proper man page, text file or so) for full options list.


-- 
Jakub Bogusz    http://qboosh.pl/


From jajcus at jajcus.net  Thu Nov  1 18:52:12 2012
From: jajcus at jajcus.net (Jacek Konieczny)
Date: Thu, 1 Nov 2012 18:52:12 +0100
Subject: systemd unit options [[packages/corosync] make
 sysconfig/corosync-notifyd usable for systemd]
In-Reply-To: <20121101131322.GB25496@mail>
References: <9c0f7a6e7d1349446f09273cc3272f77863bcf7c_refs_heads_master@pld-linux.org>
 <ac647687b5d8c7bfb01cd7fc24ef36c23b10cc23_refs_heads_master@pld-linux.org>
 <20121029163518.GA9794@mail> <20121030074023.GA1826@home.lan>
 <20121101131322.GB25496@mail>
Message-ID: <20121101175212.GC2699@lolek.nigdzie>

On Thu, Nov 01, 2012 at 02:13:22PM +0100, Jakub Bogusz wrote:
> That would be acceptable - document at least those options, which were
> supported by SysV init script and add a pointer to documentation
> (proper man page, text file or so) for full options list.

No descriptive options in the sysconfig file were supported by SysV init
script in PLD Th:

poldek -n th:

poldek:/all-avail> search -f /etc/sysconfig/corosync-notifyd
Searching packages..........................................done.
No package matches '/etc/sysconfig/corosync-notifyd'

The options you are worried about was added by me and removed by me
and never went past th-test.

Manual page reference has already been added to the sysconfig file.

Greets,
        Jacek


From qboosh at pld-linux.org  Thu Nov  1 19:16:30 2012
From: qboosh at pld-linux.org (Jakub Bogusz)
Date: Thu, 1 Nov 2012 19:16:30 +0100
Subject: systemd unit options [[packages/corosync] make
 sysconfig/corosync-notifyd usable for systemd]
In-Reply-To: <20121101175212.GC2699@lolek.nigdzie>
References: <9c0f7a6e7d1349446f09273cc3272f77863bcf7c_refs_heads_master@pld-linux.org>
 <ac647687b5d8c7bfb01cd7fc24ef36c23b10cc23_refs_heads_master@pld-linux.org>
 <20121029163518.GA9794@mail> <20121030074023.GA1826@home.lan>
 <20121101131322.GB25496@mail> <20121101175212.GC2699@lolek.nigdzie>
Message-ID: <20121101181630.GA26545@mail>

On Thu, Nov 01, 2012 at 06:52:12PM +0100, Jacek Konieczny wrote:
> On Thu, Nov 01, 2012 at 02:13:22PM +0100, Jakub Bogusz wrote:
> > That would be acceptable - document at least those options, which were
> > supported by SysV init script and add a pointer to documentation
> > (proper man page, text file or so) for full options list.
> 
> No descriptive options in the sysconfig file were supported by SysV init
> script in PLD Th:
> 
> poldek -n th:
> 
> poldek:/all-avail> search -f /etc/sysconfig/corosync-notifyd
> Searching packages..........................................done.
> No package matches '/etc/sysconfig/corosync-notifyd'
> 
> The options you are worried about was added by me and removed by me
> and never went past th-test.

That was a general remark; your change in corosync just reminded me
about similar changes in many other packages some time ago (when
I didn't have enough time to comment them).

> Manual page reference has already been added to the sysconfig file.

OK


-- 
Jakub Bogusz    http://qboosh.pl/


From glen at pld-linux.org  Thu Nov  1 19:29:37 2012
From: glen at pld-linux.org (=?ISO-8859-1?Q?Elan_Ruusam=E4e?=)
Date: Thu, 01 Nov 2012 20:29:37 +0200
Subject: [packages/filesystem] - can't use any non-root groups in this
 package and not depend on setup package, do similar workardo
In-Reply-To: <20121101131051.GA25496@mail>
References: <84956be926dc0639d1c8c3c0092fab46e402d86e_refs_heads_master@pld-linux.org>
 <aefd25979c44a66fc4eecfc32a27d2d119f31d76_refs_heads_master@pld-linux.org>
 <20121101131051.GA25496@mail>
Message-ID: <5092BF91.8090700@pld-linux.org>

On 01.11.2012 15:10, Jakub Bogusz wrote:
> On Wed, Oct 31, 2012 at 07:42:21AM +0100, glen wrote:
>> >commit aefd25979c44a66fc4eecfc32a27d2d119f31d76
>> >Author: Elan Ruusam?e<glen at delfi.ee>
>> >Date:   Wed Oct 31 08:41:13 2012 +0200
>> >
>> >     - can't use any non-root groups in this package and not depend on setup package, do similar workardound as FHS package
> And what is the problem with requiring setup here?
> setup requires just FHS, not filesystem.
>
> Your change causes rpm -V to report /var/log/archive group inconsistency on every system.
>
>
so does it in FHS, no-one complained.

we need proper fix here.
simplistic one would be if rpm allowed numeric ownership in %files 
specification

-- 
glen



From n3npq at me.com  Thu Nov  1 19:44:45 2012
From: n3npq at me.com (Jeffrey Johnson)
Date: Thu, 01 Nov 2012 14:44:45 -0400
Subject: [packages/filesystem] - can't use any non-root groups in this
 package and not depend on setup package, do similar workardo
In-Reply-To: <5092BF91.8090700@pld-linux.org>
References: <84956be926dc0639d1c8c3c0092fab46e402d86e_refs_heads_master@pld-linux.org>
 <aefd25979c44a66fc4eecfc32a27d2d119f31d76_refs_heads_master@pld-linux.org>
 <20121101131051.GA25496@mail> <5092BF91.8090700@pld-linux.org>
Message-ID: <60633CE5-9575-4E40-8BCF-3CFCEDCD1E8F@me.com>


On Nov 1, 2012, at 2:29 PM, Elan Ruusam?e wrote:

> On 01.11.2012 15:10, Jakub Bogusz wrote:
>> On Wed, Oct 31, 2012 at 07:42:21AM +0100, glen wrote:
>>> >commit aefd25979c44a66fc4eecfc32a27d2d119f31d76
>>> >Author: Elan Ruusam?e<glen at delfi.ee>
>>> >Date:   Wed Oct 31 08:41:13 2012 +0200
>>> >
>>> >     - can't use any non-root groups in this package and not depend on setup package, do similar workardound as FHS package
>> And what is the problem with requiring setup here?
>> setup requires just FHS, not filesystem.
>> 
>> Your change causes rpm -V to report /var/log/archive group inconsistency on every system.
>> 
>> 
> so does it in FHS, no-one complained.
> 
> we need proper fix here.

Yes you do.

> simplistic one would be if rpm allowed numeric ownership in %files specification
> 

Numeric was what was used in RPM in 1997, changed to use strings through /etc/passwd
lookup in 1998 (with libc5) for maximum flexibility/configurability.

Meanwhile here is what is wrong with "modern" distros
	The /etc/passwd file MUST exist for lookups to occur in packages that follow.

The original intent was setup was first pkg installed with /etc/passwd, and filesystem (or FHS or ?)
carried a prerequisite (which is Requires: in modern RPM).

The problem there is that dependency LOOP's cause setup (or whatever carries /etc/passwd)
_NOT_ to be installed first.

So there is a band-aid in RPM where certain users/groups are hardwired for
lookup. Unless additional users/groups (in this case the "logs" group) are
compiled into RPM, then there are --verify failures.

The simplest fix is to ensure that /etc/passwd (and group) is _ALWAYS_ present in empty chroot's.

The next simplest fix is to ensure that /etc/passwd (and group) ALWAYS contains all
users/groups for libc prerequisites (libc carries the name service libraries needed to do
the lookup).

There are also "probe dependencies" to ensure that user/group can be looked up _BEFORE_
any package is installed. These look like
	Requires: user(root)
or (with numeric id assertion)
	Requires: user(root) = 0
and group(?) for group lookups.

At this point in time, there is simple,y no reason _NOT_ to hardwire a check that all
user/group strings mentioned in _EVERY_ package can be looked up, with a virtual
	Provides: user(foo) = 1234
to hint that a script let is about to invoke user add etc.

Since shadowutils (yes _ALL_ of shadowutils) is carried in rpm-5.x.y through lua
bindings, there is no need to add snarly dependencies on user add packages etc.
RPM+LUA+SHADOWUTILS is every bit as functional as /usr/sbin/useradd is.

hth

73 de Jeff
> -- 
> glen
> 
> _______________________________________________
> pld-devel-en mailing list
> pld-devel-en at lists.pld-linux.org
> http://lists.pld-linux.org/mailman/listinfo/pld-devel-en



From glen at pld-linux.org  Thu Nov  1 21:02:35 2012
From: glen at pld-linux.org (=?windows-1252?Q?Elan_Ruusam=E4e?=)
Date: Thu, 01 Nov 2012 22:02:35 +0200
Subject: [packages/filesystem] - can't use any non-root groups in this
 package and not depend on setup package, do similar workardo
In-Reply-To: <60633CE5-9575-4E40-8BCF-3CFCEDCD1E8F@me.com>
References: <84956be926dc0639d1c8c3c0092fab46e402d86e_refs_heads_master@pld-linux.org>
 <aefd25979c44a66fc4eecfc32a27d2d119f31d76_refs_heads_master@pld-linux.org>
 <20121101131051.GA25496@mail> <5092BF91.8090700@pld-linux.org>
 <60633CE5-9575-4E40-8BCF-3CFCEDCD1E8F@me.com>
Message-ID: <5092D55B.4080603@pld-linux.org>

On 11/01/2012 08:44 PM, Jeffrey Johnson wrote:
> The original intent was setup was first pkg installed with /etc/passwd, and filesystem (or FHS or ?)
> carried a prerequisite (which is Requires: in modern RPM).
that's the thing, currently FHS (or filesystem) needs to be installed 
before setup package, due /etc dir-dependency :)

-- 
glen



From n3npq at me.com  Thu Nov  1 21:08:23 2012
From: n3npq at me.com (Jeffrey Johnson)
Date: Thu, 01 Nov 2012 16:08:23 -0400
Subject: [packages/filesystem] - can't use any non-root groups in this
 package and not depend on setup package, do similar workardo
In-Reply-To: <5092D55B.4080603@pld-linux.org>
References: <84956be926dc0639d1c8c3c0092fab46e402d86e_refs_heads_master@pld-linux.org>
 <aefd25979c44a66fc4eecfc32a27d2d119f31d76_refs_heads_master@pld-linux.org>
 <20121101131051.GA25496@mail> <5092BF91.8090700@pld-linux.org>
 <60633CE5-9575-4E40-8BCF-3CFCEDCD1E8F@me.com> <5092D55B.4080603@pld-linux.org>
Message-ID: <90032C14-1C63-4C70-B9DC-BDB85025C196@me.com>


On Nov 1, 2012, at 4:02 PM, Elan Ruusam?e wrote:

> On 11/01/2012 08:44 PM, Jeffrey Johnson wrote:
>> The original intent was setup was first pkg installed with /etc/passwd, and filesystem (or FHS or ?)
>> carried a prerequisite (which is Requires: in modern RPM).
> that's the thing, currently FHS (or filesystem) needs to be installed before setup package, due /etc dir-dependency :)
> 


Yep: a dependency LOOP with all edges in the LOOP ignored for ordering purposes.

So put /etc/passwd (and group) into FHS ? the only requirement is that /etc/passwd
(and group) must be installed before any package that uses user:group other than root:root
(or what is hardwired in RPM itself).

Alternatively, put /etc (and /) into the setup package so that the parent-dir sep is resolved.

There shouldn't be a problem if multiple packages own a single directory: rmdir(2) fails on
non-empty directories and errno is ignored.

73 de Jeff



From gotar at polanet.pl  Thu Nov  1 21:16:40 2012
From: gotar at polanet.pl (Tomasz Pala)
Date: Thu, 1 Nov 2012 21:16:40 +0100
Subject: [packages/filesystem] - can't use any non-root groups in this
 package and not depend on setup package, do similar workardo
In-Reply-To: <5092D55B.4080603@pld-linux.org>
References: <84956be926dc0639d1c8c3c0092fab46e402d86e_refs_heads_master@pld-linux.org>
 <aefd25979c44a66fc4eecfc32a27d2d119f31d76_refs_heads_master@pld-linux.org>
 <20121101131051.GA25496@mail> <5092BF91.8090700@pld-linux.org>
 <60633CE5-9575-4E40-8BCF-3CFCEDCD1E8F@me.com>
 <5092D55B.4080603@pld-linux.org>
Message-ID: <20121101201639.GA7581@polanet.pl>

On Thu, Nov 01, 2012 at 22:02:35 +0200, Elan Ruusam?e wrote:

>> The original intent was setup was first pkg installed with /etc/passwd, and filesystem (or FHS or ?)
>> carried a prerequisite (which is Requires: in modern RPM).
> that's the thing, currently FHS (or filesystem) needs to be installed 
> before setup package, due /etc dir-dependency :)

PLD-like solution follows:
1. FHS-etc.rpm carrying /etc directory only
2. setup-passwd.rpm carrying /etc/passwd with R: FHS-etc
3. FHS.rpm carrying the rest of FHS with R: setup-passwd
4. setup.rpm containing the rest of current setup package, R: FHS (no changes here).

Chicken-egg problem solved! Thanks again, subpackages!

-- 
Tomasz Pala <gotar at pld-linux.org>


From n3npq at me.com  Thu Nov  1 21:22:28 2012
From: n3npq at me.com (Jeffrey Johnson)
Date: Thu, 01 Nov 2012 16:22:28 -0400
Subject: [packages/filesystem] - can't use any non-root groups in this
 package and not depend on setup package, do similar workardo
In-Reply-To: <20121101201639.GA7581@polanet.pl>
References: <84956be926dc0639d1c8c3c0092fab46e402d86e_refs_heads_master@pld-linux.org>
 <aefd25979c44a66fc4eecfc32a27d2d119f31d76_refs_heads_master@pld-linux.org>
 <20121101131051.GA25496@mail> <5092BF91.8090700@pld-linux.org>
 <60633CE5-9575-4E40-8BCF-3CFCEDCD1E8F@me.com> <5092D55B.4080603@pld-linux.org>
 <20121101201639.GA7581@polanet.pl>
Message-ID: <2A0D6DD8-4808-4A0E-8FC4-866FC2E48C04@me.com>


On Nov 1, 2012, at 4:16 PM, Tomasz Pala wrote:

> On Thu, Nov 01, 2012 at 22:02:35 +0200, Elan Ruusam?e wrote:
> 
>>> The original intent was setup was first pkg installed with /etc/passwd, and filesystem (or FHS or ?)
>>> carried a prerequisite (which is Requires: in modern RPM).
>> that's the thing, currently FHS (or filesystem) needs to be installed 
>> before setup package, due /etc dir-dependency :)
> 
> PLD-like solution follows:
> 1. FHS-etc.rpm carrying /etc directory only
> 2. setup-passwd.rpm carrying /etc/passwd with R: FHS-etc
> 3. FHS.rpm carrying the rest of FHS with R: setup-passwd
> 4. setup.rpm containing the rest of current setup package, R: FHS (no changes here).
> 
> Chicken-egg problem solved! Thanks again, sub packages!

Well yes, that's a solution too.

Lots of teensy packages with single directory/file
just to satisfy dependencies has its own complexity costs.
73 de Jeff



From jajcus at jajcus.net  Sun Nov  4 11:41:25 2012
From: jajcus at jajcus.net (Jacek Konieczny)
Date: Sun, 4 Nov 2012 11:41:25 +0100
Subject: RFC: EFI System partition and native EFI binaries (xen.efi,
 vmlinuz.efi, etc)
In-Reply-To: <20121029125039.GG8415@jajo.eggsoft>
References: <20121029125039.GG8415@jajo.eggsoft>
Message-ID: <20121104104124.GA2162@lolek.nigdzie>

On Mon, Oct 29, 2012 at 01:50:39PM +0100, Jacek Konieczny wrote:
> Any thoughts? Does anybody know how other distros solve this problem?

I have not expected such a heated discussion here? thanks for all your
comments and suggestions?

So I have created another monster ? the 'efi-boot-update' script.
Packages providing EFI boot loaders install them to the
/lib/efi/{ia32,x64} directories and their configuration files to the
/etc/efi-boot/ directory. Additionally, an efi-boot-update configuration
files are provided in the /etc/efi-boot/update.d directory which
describe what files should be copied to the EFI partition for each boot
loader and what boot manager label to use.

/etc/efi-boot/update.conf file is used to define the boot order and
other global options for the efi-boot-update scripts. Everything seems
to work good enough for me. It would be great if anybody else could test
this too and tell if it makes sense.

The packages supporting efi-boot-update now:
? xen-efi 4.2.0-1 (xen.spec)
? elilo 3.14-2 (elilo.spec)
? efi-shell-{ia32,x64) 2.0 (efi-shell.spec)

Additionally, the efi-boot-update script contains a configuration file
to set a boot manager entry for GRUB2 installed as usual with
'grub-install'.

TODO:
- if this really makes sense, then the /lib/efi/{ia32,x64} and
  /etc/efi-boot/  directories should be probably moved to the filesystem
  package.

I will also try to write wiki page about EFI boot.

Greets,
        Jacek


From qboosh at pld-linux.org  Sun Nov  4 11:52:11 2012
From: qboosh at pld-linux.org (Jakub Bogusz)
Date: Sun, 4 Nov 2012 11:52:11 +0100
Subject: [packages/efi-shell/notes/commits] Notes added by 'git notes add'
In-Reply-To: <2ae807df00a23e6b1b9823da11c1cc38231b5ae0_refs_notes_commits@pld-linux.org>
References: <20121104104103.12024.70479@pld-linux.org>
 <2ae807df00a23e6b1b9823da11c1cc38231b5ae0_refs_notes_commits@pld-linux.org>
Message-ID: <20121104105211.GA11990@mail>

Note: with current git log -> %changelog generation you can't begin
changelog entry with "%post" (or other rpm section keyword) - it would
cause changelog to end immediately, thus leading to spec syntax error.

That's why I added git note overriding changelog entry for this commit.


On Sun, Nov 04, 2012 at 11:41:08AM +0100, qboosh wrote:
> commit 2ae807df00a23e6b1b9823da11c1cc38231b5ae0
> Author: Jakub Bogusz <qboosh at pld-linux.org>
> Date:   Sun Nov 4 11:38:18 2012 +0100
> 
>     Notes added by 'git notes add'
> 
>  af5feaae61966c7e64cdf39baececae4c4f5b0f4 | 1 +
>  1 file changed, 1 insertion(+)
> ---
> diff --git a/af5feaae61966c7e64cdf39baececae4c4f5b0f4 b/af5feaae61966c7e64cdf39baececae4c4f5b0f4
> new file mode 100644
> index 0000000..8cb4260
> --- /dev/null
> +++ b/af5feaae61966c7e64cdf39baececae4c4f5b0f4
> @@ -0,0 +1 @@
> +- %post improved, %postun added
> ================================================================
> 
> ---- gitweb:
> 
> http://git.pld-linux.org/gitweb.cgi/packages/efi-shell.git/commitdiff/2ae807df00a23e6b1b9823da11c1cc38231b5ae0
> 
> _______________________________________________
> pld-cvs-commit mailing list
> pld-cvs-commit at lists.pld-linux.org
> http://lists.pld-linux.org/mailman/listinfo/pld-cvs-commit

-- 
Jakub Bogusz    http://qboosh.pl/


From n3npq at me.com  Sun Nov  4 20:37:48 2012
From: n3npq at me.com (Jeffrey Johnson)
Date: Sun, 04 Nov 2012 14:37:48 -0500
Subject: [packages/efi-shell/notes/commits] Notes added by 'git notes add'
In-Reply-To: <20121104105211.GA11990@mail>
References: <20121104104103.12024.70479@pld-linux.org>
 <2ae807df00a23e6b1b9823da11c1cc38231b5ae0_refs_notes_commits@pld-linux.org>
 <20121104105211.GA11990@mail>
Message-ID: <764BE44E-F325-4192-B95F-A583D235CCC6@me.com>


On Nov 4, 2012, at 5:52 AM, Jakub Bogusz wrote:

> Note: with current git log -> %changelog generation you can't begin
> changelog entry with "%post" (or other rpm section keyword) - it would
> cause changelog to end immediately, thus leading to spec syntax error.
> 

The problem is actually worse if/when %post (as a section marker)
is overloaded as %poist (as a macro), particularly with multiple lines.

I'd suggest %%post escaping ? ymmv.



> That's why I added git note overriding changelog entry for this commit.
> 
> 
> On Sun, Nov 04, 2012 at 11:41:08AM +0100, qboosh wrote:
>> commit 2ae807df00a23e6b1b9823da11c1cc38231b5ae0
>> Author: Jakub Bogusz <qboosh at pld-linux.org>
>> Date:   Sun Nov 4 11:38:18 2012 +0100
>> 
>>    Notes added by 'git notes add'
>> 
>> af5feaae61966c7e64cdf39baececae4c4f5b0f4 | 1 +
>> 1 file changed, 1 insertion(+)
>> ---
>> diff --git a/af5feaae61966c7e64cdf39baececae4c4f5b0f4 b/af5feaae61966c7e64cdf39baececae4c4f5b0f4
>> new file mode 100644
>> index 0000000..8cb4260
>> --- /dev/null
>> +++ b/af5feaae61966c7e64cdf39baececae4c4f5b0f4
>> @@ -0,0 +1 @@
>> +- %post improved, %postun added
>> ================================================================
>> 
>> ---- gitweb:
>> 
>> http://git.pld-linux.org/gitweb.cgi/packages/efi-shell.git/commitdiff/2ae807df00a23e6b1b9823da11c1cc38231b5ae0
>> 
>> _______________________________________________
>> pld-cvs-commit mailing list
>> pld-cvs-commit at lists.pld-linux.org
>> http://lists.pld-linux.org/mailman/listinfo/pld-cvs-commit
> 
> -- 
> Jakub Bogusz    http://qboosh.pl/
> _______________________________________________
> pld-devel-en mailing list
> pld-devel-en at lists.pld-linux.org
> http://lists.pld-linux.org/mailman/listinfo/pld-devel-en



From glen at pld-linux.org  Sun Nov  4 22:10:56 2012
From: glen at pld-linux.org (=?ISO-8859-1?Q?Elan_Ruusam=E4e?=)
Date: Sun, 04 Nov 2012 23:10:56 +0200
Subject: [packages/efi-shell/notes/commits] Notes added by 'git notes add'
In-Reply-To: <20121104105211.GA11990@mail>
References: <20121104104103.12024.70479@pld-linux.org>
 <2ae807df00a23e6b1b9823da11c1cc38231b5ae0_refs_notes_commits@pld-linux.org>
 <20121104105211.GA11990@mail>
Message-ID: <5096D9E0.6080707@pld-linux.org>

On 11/04/2012 12:52 PM, Jakub Bogusz wrote:
> Note: with current git log -> %changelog generation you can't begin
> changelog entry with "%post" (or other rpm section keyword) - it would
> cause changelog to end immediately, thus leading to spec syntax error.
>
> That's why I added git note overriding changelog entry for this commit.
>
are you sure it's from that, that it's %changelog  generated from git log?

perhaps rpm5 yet again parses %changelog for macros, in rpm45 we had 
patch to skip any parsing there.

-- 
glen



From jajcus at jajcus.net  Sun Nov  4 22:20:18 2012
From: jajcus at jajcus.net (Jacek Konieczny)
Date: Sun, 4 Nov 2012 22:20:18 +0100
Subject: [packages/efi-shell/notes/commits] Notes added by 'git notes add'
In-Reply-To: <5096D9E0.6080707@pld-linux.org>
References: <20121104104103.12024.70479@pld-linux.org>
 <2ae807df00a23e6b1b9823da11c1cc38231b5ae0_refs_notes_commits@pld-linux.org>
 <20121104105211.GA11990@mail> <5096D9E0.6080707@pld-linux.org>
Message-ID: <20121104212017.GB2162@lolek.nigdzie>

On Sun, Nov 04, 2012 at 11:10:56PM +0200, Elan Ruusam?e wrote:
> On 11/04/2012 12:52 PM, Jakub Bogusz wrote:
> > Note: with current git log -> %changelog generation you can't begin
> > changelog entry with "%post" (or other rpm section keyword) - it would
> > cause changelog to end immediately, thus leading to spec syntax error.
> >
> > That's why I added git note overriding changelog entry for this commit.
> >
> are you sure it's from that, that it's %changelog  generated from git log?
> 
> perhaps rpm5 yet again parses %changelog for macros, in rpm45 we had 
> patch to skip any parsing there.

I have the same problem with rpm 4.5.

Greets,
        Jacek


From glen at pld-linux.org  Sun Nov  4 23:07:03 2012
From: glen at pld-linux.org (=?ISO-8859-1?Q?Elan_Ruusam=E4e?=)
Date: Mon, 05 Nov 2012 00:07:03 +0200
Subject: [packages/efi-shell/notes/commits] Notes added by 'git notes add'
In-Reply-To: <20121104212017.GB2162@lolek.nigdzie>
References: <20121104104103.12024.70479@pld-linux.org>
 <2ae807df00a23e6b1b9823da11c1cc38231b5ae0_refs_notes_commits@pld-linux.org>
 <20121104105211.GA11990@mail> <5096D9E0.6080707@pld-linux.org>
 <20121104212017.GB2162@lolek.nigdzie>
Message-ID: <5096E707.7070501@pld-linux.org>

On 11/04/2012 11:20 PM, Jacek Konieczny wrote:
> On Sun, Nov 04, 2012 at 11:10:56PM +0200, Elan Ruusam?e wrote:
>> >On 11/04/2012 12:52 PM, Jakub Bogusz wrote:
>>> > >Note: with current git log -> %changelog generation you can't begin
>>> > >changelog entry with "%post" (or other rpm section keyword) - it would
>>> > >cause changelog to end immediately, thus leading to spec syntax error.
>>> > >
>>> > >That's why I added git note overriding changelog entry for this commit.
>>> > >
>> >are you sure it's from that, that it's %changelog  generated from git log?
>> >
>> >perhaps rpm5 yet again parses %changelog for macros, in rpm45 we had
>> >patch to skip any parsing there.
> I have the same problem with rpm 4.5.
ah, indeed, now i "see"

%post in the beginning of line, starts that section, regardless that we 
are "in" %changelog section
in pld, as %changelog is last section, could perhaps omit parsing that 
as well? :)

altho it's farily easy to replace ^% -> ^%% in the generated changelog, 
but then, when we "fix" rpmbuild side, it won't be consistent anymore.

so how we fix it? :)

-- 
glen



From mike at osdn.org.ua  Sun Nov  4 23:10:49 2012
From: mike at osdn.org.ua (Michael Shigorin)
Date: Mon, 5 Nov 2012 00:10:49 +0200
Subject: [packages/efi-shell/notes/commits] Notes added by 'git notes add'
In-Reply-To: <5096E707.7070501@pld-linux.org>
References: <20121104104103.12024.70479@pld-linux.org>
 <2ae807df00a23e6b1b9823da11c1cc38231b5ae0_refs_notes_commits@pld-linux.org>
 <20121104105211.GA11990@mail> <5096D9E0.6080707@pld-linux.org>
 <20121104212017.GB2162@lolek.nigdzie> <5096E707.7070501@pld-linux.org>
Message-ID: <20121104221049.GD11474@osdn.org.ua>

On Mon, Nov 05, 2012 at 12:07:03AM +0200, Elan Ruusam?e wrote:
> so how we fix it? :)

In ALT Linux we've agreed to escape section names
(and most of the macros) in %changelog :)

-- 
my two copecks


From n3npq at mac.com  Mon Nov  5 04:18:50 2012
From: n3npq at mac.com (Jeffrey Johnson)
Date: Sun, 04 Nov 2012 22:18:50 -0500
Subject: [packages/efi-shell/notes/commits] Notes added by 'git notes add'
In-Reply-To: <5096E707.7070501@pld-linux.org>
References: <20121104104103.12024.70479@pld-linux.org>
 <2ae807df00a23e6b1b9823da11c1cc38231b5ae0_refs_notes_commits@pld-linux.org>
 <20121104105211.GA11990@mail> <5096D9E0.6080707@pld-linux.org>
 <20121104212017.GB2162@lolek.nigdzie> <5096E707.7070501@pld-linux.org>
Message-ID: <740121BF-E293-4BE0-A73B-CA2F51B9B75D@mac.com>




> 
> so how we fix it

There isn't a simple fix.

Macro's are expanded when encountered as the simplest
possible (and most general)
solution.

It's certainly possible to make
the *.spec parser section dependent.

However there are many problematic
constricts if that path is pursued.

hth

73 de Jeff


From draenog at pld-linux.org  Mon Nov  5 13:05:22 2012
From: draenog at pld-linux.org (Kacper Kornet)
Date: Mon, 5 Nov 2012 13:05:22 +0100
Subject: Empty packages
Message-ID: <20121105120521.GA21759@camk.edu.pl>

The following packages have been created but nothing has ever been
pushed to them:

drush.git
gstreamer1.git
keepassx.git
perl-Net-UPnP.git
perl-RT-Extension-JSGantt.git

Unless someone protests they will be removed.

-- 
  Kacper


From qboosh at pld-linux.org  Sat Nov 10 15:14:00 2012
From: qboosh at pld-linux.org (Jakub Bogusz)
Date: Sat, 10 Nov 2012 15:14:00 +0100
Subject: texlive errors [builder-th-i486@pld-linux.org: TEST build ERRORS:
 nickle.spec OK: ibus-gjs.spec]
Message-ID: <20121110141400.GA6044@mail>

Any hints what's wrong?
Google answers are very unclear.

It failed on th-i[46]86, built successfully on th-x86_64 and my i686
machine.

----- Forwarded message from PLD th-i486 builder <builder-th-i486 at pld-linux.org> -----

[...]
docbook2pdf nickle-tutorial.sgml
Using catalogs: /etc/sgml/catalog
Using stylesheet: /usr/share/sgml/docbook/utils-0.6.14/docbook-utils.dsl#print
Working on: /tmp/B.986ad46e-05bc-4d8a-984a-56321f8fa400/BUILD/nickle-2.77/doc/tutorial/nickle-tutorial.sgml
This is pdfTeXk, Version 3.1415926-1.40.9 (Web2C 7.5.7)
 %&-line parsing enabled.
entering extended mode

(/tmp/B.986ad46e-05bc-4d8a-984a-56321f8fa400/BUILD/nickle-2.77/doc/tutorial/nic
kle-tutorial.tex
JadeTeX 2002/01/25: 3.12
(/usr/share/texmf-dist/tex/latex/psnfss/t1ptm.fd)
Elements will be labelled
Jade begin document sequence at 21
No file nickle-tutorial.aux.
(/usr/share/texmf-dist/tex/latex/tipa/t3ptm.fd)
No file T2Acmr.fd.

! LaTeX Error: This NFSS system isn't set up properly.

See the LaTeX manual or LaTeX Companion for explanation.
Type  H <return>  for immediate help.
 ...                                              
                                                  
[...]
Transcript written on nickle-tutorial.log.
make[3]: *** [nickle-tutorial.pdf] Error 9
make[3]: Leaving directory `/tmp/B.986ad46e-05bc-4d8a-984a-56321f8fa400/BUILD/nickle-2.77/doc/tutorial'
make[2]: *** [all-recursive] Error 1
make[2]: Leaving directory `/tmp/B.986ad46e-05bc-4d8a-984a-56321f8fa400/BUILD/nickle-2.77/doc'
make[1]: *** [all-recursive] Error 1
make[1]: Leaving directory `/tmp/B.986ad46e-05bc-4d8a-984a-56321f8fa400/BUILD/nickle-2.77'
make: *** [all] Error 2
error: Bad exit status from /tmp/B.986ad46e-05bc-4d8a-984a-56321f8fa400/BUILD/tmp/rpm-tmp.66362 (%build)


RPM build errors:
    Bad exit status from /tmp/B.986ad46e-05bc-4d8a-984a-56321f8fa400/BUILD/tmp/rpm-tmp.66362 (%build)
ended at: Sat Nov 10 12:10:01 2012, done in 0:00:09.842846
error: No files produced.
+ chmod -R u+rwX /tmp/B.986ad46e-05bc-4d8a-984a-56321f8fa400/BUILD
+ rm -rf /tmp/B.986ad46e-05bc-4d8a-984a-56321f8fa400/tmp /tmp/B.986ad46e-05bc-4d8a-984a-56321f8fa400/BUILD
+ rm -rf /tmp/B.986ad46e-05bc-4d8a-984a-56321f8fa400
Begin-PLD-Builder-Info
Build-Time: user:8.11s sys:1.22s real:14.58s (faults io:35 non-io:786808)

End-PLD-Builder-Info


----- End forwarded message -----

-- 
Jakub Bogusz    http://qboosh.pl/


From blues at pld-linux.org  Sun Nov 11 12:56:34 2012
From: blues at pld-linux.org (=?ISO-8859-2?Q?Pawe=B3_Go=B3aszewski?=)
Date: Sun, 11 Nov 2012 12:56:34 +0100 (CET)
Subject: [packages/lighttpd] the systemd service file actually added
In-Reply-To: <b4a3acf23fb992d9e8fb7a3bb5a415bc535c1f9b_refs_heads_master@pld-linux.org>
References: <52e7e8c49b75b385c0eff911eec5b835c1d7a771_refs_heads_master@pld-linux.org>
 <b4a3acf23fb992d9e8fb7a3bb5a415bc535c1f9b_refs_heads_master@pld-linux.org>
Message-ID: <alpine.LNX.2.00.1211111255140.5999@fgnss.gnfx.tqn.cy>

On Sun, 11 Nov 2012, jajcus wrote:
> commit b4a3acf23fb992d9e8fb7a3bb5a415bc535c1f9b
> Author: Jacek Konieczny <jajcus at jajcus.net>
> Date:   Sun Nov 11 12:26:05 2012 +0100
> 
>     the systemd service file actually added
[...]
> +[Service]
> +EnvironmentFile=-/etc/sysconfig/lighttpd
> +ExecStartPre=/usr/sbin/lighttpd-angel -t -f /etc/lighttpd/lighttpd.conf
> +ExecStart=/usr/sbin/lighttpd-angel -D -f /etc/lighttpd/lighttpd.conf $HTTPD_OPTS
> +ExecReload=/bin/kill -HUP $MAINPID

Co? mi si? wydaje, ?e w przypadku systemd angel jest kompletnie zb?dnym 
ficzerem i sam systemd za?atwi spraw?..

-- 
pozdr.  Pawe? Go?aszewski          jid:blues<at>jabber<dot>gda<dot>pl
--------------------------------------------------------------------------
If you think of MS-DOS as mono, and Windows as stereo, then Linux is Dolby
Pro-Logic Surround Sound with Bass Boost and all the music is free.

From jajcus at jajcus.net  Sun Nov 11 16:58:01 2012
From: jajcus at jajcus.net (Jacek Konieczny)
Date: Sun, 11 Nov 2012 16:58:01 +0100
Subject: error: File(s) packaged into both ...
Message-ID: <20121111155801.GF2095@lolek.nigdzie>

Hello,

Since when is this an error? RPM5?

> error: File(s) packaged into both bacula-console-wx-5.2.12-1.x86_64 and bacula-console-qt4-5.2.12-1.x86_64:
>        /usr/share/pixmaps/bacula.png

from:
http://buildlogs.pld-linux.org/index.php?dist=th&arch=x86_64&ok=0&name=bacula&id=9c2d1726-4c45-4a22-8b84-2d616936e040&action=tail

It was often very handy to package a file into several packages ? no
need to create a 'common' subpackage every time a few, possibly
alternative, sub-packages required the same file.

Is such behaviour an error now? Or have I missed some more serious problem?

Greets,
        Jacek


From qboosh at pld-linux.org  Sun Nov 11 17:07:29 2012
From: qboosh at pld-linux.org (Jakub Bogusz)
Date: Sun, 11 Nov 2012 17:07:29 +0100
Subject: error: File(s) packaged into both ...
In-Reply-To: <20121111155801.GF2095@lolek.nigdzie>
References: <20121111155801.GF2095@lolek.nigdzie>
Message-ID: <20121111160729.GA12762@mail>

On Sun, Nov 11, 2012 at 04:58:01PM +0100, Jacek Konieczny wrote:
> Hello,
> 
> Since when is this an error? RPM5?
> 
> > error: File(s) packaged into both bacula-console-wx-5.2.12-1.x86_64 and bacula-console-qt4-5.2.12-1.x86_64:
> >        /usr/share/pixmaps/bacula.png
> 
> from:
> http://buildlogs.pld-linux.org/index.php?dist=th&arch=x86_64&ok=0&name=bacula&id=9c2d1726-4c45-4a22-8b84-2d616936e040&action=tail
> 
> It was often very handy to package a file into several packages ??? no
> need to create a 'common' subpackage every time a few, possibly
> alternative, sub-packages required the same file.
> 
> Is such behaviour an error now? Or have I missed some more serious problem?

Since RPM5. It's configurable, the default is an error in PLD.
Probably can be disabled in spec.


-- 
Jakub Bogusz    http://qboosh.pl/


From n3npq at me.com  Sun Nov 11 22:06:50 2012
From: n3npq at me.com (Jeffrey Johnson)
Date: Sun, 11 Nov 2012 16:06:50 -0500
Subject: error: File(s) packaged into both ...
In-Reply-To: <20121111160729.GA12762@mail>
References: <20121111155801.GF2095@lolek.nigdzie> <20121111160729.GA12762@mail>
Message-ID: <7800D1E6-A0C6-406A-B95C-F8A5D0882435@me.com>


On Nov 11, 2012, at 11:07 AM, Jakub Bogusz <qboosh at pld-linux.org> wrote:

> On Sun, Nov 11, 2012 at 04:58:01PM +0100, Jacek Konieczny wrote:
>> Hello,
>> 
>> Since when is this an error? RPM5?
>> 
>>> error: File(s) packaged into both bacula-console-wx-5.2.12-1.x86_64 and bacula-console-qt4-5.2.12-1.x86_64:
>>>       /usr/share/pixmaps/bacula.png
>> 
>> from:
>> http://buildlogs.pld-linux.org/index.php?dist=th&arch=x86_64&ok=0&name=bacula&id=9c2d1726-4c45-4a22-8b84-2d616936e040&action=tail
>> 
>> It was often very handy to package a file into several packages ??? no
>> need to create a 'common' subpackage every time a few, possibly
>> alternative, sub-packages required the same file.
>> 
>> Is such behaviour an error now? Or have I missed some more serious problem?
> 
> Since RPM5. It's configurable, the default is an error in PLD.
> Probably can be disabled in spec.
> 

Yes.

Note that having a file "replace" another package's content
on same path has been ripped out @rpm5.org.

Never worked well, mostly obscure functionality, and quite
costly to compute are the reasons for the removal.

73 de Jeff


From mike at osdn.org.ua  Sun Nov 11 22:20:19 2012
From: mike at osdn.org.ua (Michael Shigorin)
Date: Sun, 11 Nov 2012 23:20:19 +0200
Subject: error: File(s) packaged into both ...
In-Reply-To: <7800D1E6-A0C6-406A-B95C-F8A5D0882435@me.com>
References: <20121111155801.GF2095@lolek.nigdzie>
 <20121111160729.GA12762@mail> <7800D1E6-A0C6-406A-B95C-F8A5D0882435@me.com>
Message-ID: <20121111212019.GG19661@osdn.org.ua>

On Sun, Nov 11, 2012 at 04:06:50PM -0500, Jeffrey Johnson wrote:
> Note that having a file "replace" another package's content
> on same path has been ripped out @rpm5.org.
> 
> Never worked well, mostly obscure functionality, and quite
> costly to compute are the reasons for the removal.

Argh, I've used that rarely but IIRC for a reason with ALT-RPM.

-- 
 ---- WBR, Michael Shigorin <mike at altlinux.ru>
  ------ Linux.Kiev http://www.linux.kiev.ua/


From n3npq at me.com  Sun Nov 11 22:22:07 2012
From: n3npq at me.com (Jeffrey Johnson)
Date: Sun, 11 Nov 2012 16:22:07 -0500
Subject: error: File(s) packaged into both ...
In-Reply-To: <20121111212019.GG19661@osdn.org.ua>
References: <20121111155801.GF2095@lolek.nigdzie> <20121111160729.GA12762@mail>
 <7800D1E6-A0C6-406A-B95C-F8A5D0882435@me.com>
 <20121111212019.GG19661@osdn.org.ua>
Message-ID: <E84BBD12-A733-4CA0-88D6-E04925710DF9@me.com>


On Nov 11, 2012, at 4:20 PM, Michael Shigorin <mike at osdn.org.ua> wrote:

> On Sun, Nov 11, 2012 at 04:06:50PM -0500, Jeffrey Johnson wrote:
>> Note that having a file "replace" another package's content
>> on same path has been ripped out @rpm5.org.
>> 
>> Never worked well, mostly obscure functionality, and quite
>> costly to compute are the reasons for the removal.
> 
> Argh, I've used that rarely but IIRC for a reason with ALT-RPM.
> 

Too bad for you.

Note "quite costly to compute" means exactly that.

73 de Jeff



From jajcus at jajcus.net  Sun Nov 11 22:42:49 2012
From: jajcus at jajcus.net (Jacek Konieczny)
Date: Sun, 11 Nov 2012 22:42:49 +0100
Subject: error: File(s) packaged into both ...
In-Reply-To: <7800D1E6-A0C6-406A-B95C-F8A5D0882435@me.com>
References: <20121111155801.GF2095@lolek.nigdzie> <20121111160729.GA12762@mail>
 <7800D1E6-A0C6-406A-B95C-F8A5D0882435@me.com>
Message-ID: <20121111214249.GG2095@lolek.nigdzie>

On Sun, Nov 11, 2012 at 04:06:50PM -0500, Jeffrey Johnson wrote:
> Note that having a file "replace" another package's content
> on same path

What do you mean by that? Installing two packages containing the same
path with '--force'?

Greets,
        Jacek


From n3npq at me.com  Sun Nov 11 22:46:41 2012
From: n3npq at me.com (Jeffrey Johnson)
Date: Sun, 11 Nov 2012 16:46:41 -0500
Subject: error: File(s) packaged into both ...
In-Reply-To: <20121111214249.GG2095@lolek.nigdzie>
References: <20121111155801.GF2095@lolek.nigdzie> <20121111160729.GA12762@mail>
 <7800D1E6-A0C6-406A-B95C-F8A5D0882435@me.com>
 <20121111214249.GG2095@lolek.nigdzie>
Message-ID: <D7D9B865-BE5F-4D62-81C9-F1C0D129547C@me.com>


On Nov 11, 2012, at 4:42 PM, Jacek Konieczny <jajcus at jajcus.net> wrote:

> On Sun, Nov 11, 2012 at 04:06:50PM -0500, Jeffrey Johnson wrote:
>> Note that having a file "replace" another package's content
>> on same path
> 
> What do you mean by that? Installing two packages containing the same
> path with '--force'?
> 

Nope. For starters --force isn't any one functionality.

If you've never queried a file state, then you simply will not care.

Even if you *have* queried a file state, using alternatives rather
than obscure rpm functionality that has never worked quite right
is likely saner.

73 de Jeff


From jajcus at jajcus.net  Tue Nov 13 12:25:01 2012
From: jajcus at jajcus.net (Jacek Konieczny)
Date: Tue, 13 Nov 2012 12:25:01 +0100
Subject: [kernel] 'perf' utility and other utils from kernel sources
Message-ID: <20121113112501.GC7561@jajo.eggsoft>

Hi,

I have just found this:
http://serverfault.com/questions/448125/linux-finding-the-system-processes-which-are-at-the-top
and this:
https://perf.wiki.kernel.org/index.php/Tutorial

It seems the 'perf' utility may be a great tool? but I cannot find it in
PLD. It should be probably built from kernel.spec.

Another utility in kernel sources, that I find useful is gen_init_cpio.

Should we build some 'kernel-tools'? Do we already have such thing? 

Greets,
        Jacek


From glen at pld-linux.org  Tue Nov 13 12:30:02 2012
From: glen at pld-linux.org (=?UTF-8?B?RWxhbiBSdXVzYW3DpGU=?=)
Date: Tue, 13 Nov 2012 13:30:02 +0200
Subject: [kernel] 'perf' utility and other utils from kernel sources
In-Reply-To: <20121113112501.GC7561@jajo.eggsoft>
References: <20121113112501.GC7561@jajo.eggsoft>
Message-ID: <50A22F3A.3020302@pld-linux.org>

On 13.11.2012 13:25, Jacek Konieczny wrote:
> Hi,
>
> I have just found this:
> http://serverfault.com/questions/448125/linux-finding-the-system-processes-which-are-at-the-top
> and this:
> https://perf.wiki.kernel.org/index.php/Tutorial
>
> It seems the 'perf' utility may be a great tool? but I cannot find it in
> PLD. It should be probably built from kernel.spec.
>
> Another utility in kernel sources, that I find useful is gen_init_cpio.
>
> Should we build some 'kernel-tools'? Do we already have such thing?

had the same question in irc, i was more interested in cpufreq tools

--- Day changed N nov   08 2012
07:37  glen> baggins: %package tools in kernel.spec or kernel-tools.spec ?
07:37  glen> as for example linux-libc-headers we have separate .spec, 
makes sense to keep tools also separate



-- 
glen



From arekm at maven.pl  Tue Nov 13 12:34:06 2012
From: arekm at maven.pl (Arkadiusz =?utf-8?q?Mi=C5=9Bkiewicz?=)
Date: Tue, 13 Nov 2012 12:34:06 +0100
Subject: [kernel] 'perf' utility and other utils from kernel sources
In-Reply-To: <20121113112501.GC7561@jajo.eggsoft>
References: <20121113112501.GC7561@jajo.eggsoft>
Message-ID: <201211131234.06853.arekm@maven.pl>

On Tuesday 13 of November 2012, Jacek Konieczny wrote:
> Hi,
> 
> I have just found this:
> http://serverfault.com/questions/448125/linux-finding-the-system-processes-
> which-are-at-the-top and this:
> https://perf.wiki.kernel.org/index.php/Tutorial
> 
> It seems the 'perf' utility may be a great tool? but I cannot find it in
> PLD. It should be probably built from kernel.spec.

If it's stricly dependant on kernel version then most likely should build from 
kernel.spec (and afaik it is dependant).

> Another utility in kernel sources, that I find useful is gen_init_cpio.

I guess most of tools/* should be packaged, too.

> Should we build some 'kernel-tools'? Do we already have such thing?

We don't have it.

> Greets,
>         Jacek

-- 
Arkadiusz Mi?kiewicz, arekm / maven.pl


From gotar at polanet.pl  Tue Nov 13 19:31:05 2012
From: gotar at polanet.pl (Tomasz Pala)
Date: Tue, 13 Nov 2012 19:31:05 +0100
Subject: [kernel] 'perf' utility and other utils from kernel sources
In-Reply-To: <201211131234.06853.arekm@maven.pl>
References: <20121113112501.GC7561@jajo.eggsoft>
 <201211131234.06853.arekm@maven.pl>
Message-ID: <20121113183105.GA24118@polanet.pl>

On Tue, Nov 13, 2012 at 12:34:06 +0100, Arkadiusz Mi?kiewicz wrote:

>> It seems the 'perf' utility may be a great tool??? but I cannot find it in
>> PLD. It should be probably built from kernel.spec.
> 
> If it's stricly dependant on kernel version then most likely should build from 
> kernel.spec (and afaik it is dependant).

It makes use of debug symbols, but it's not _strictly_ dependant. I use
some older build for quite a long time.

-- 
Tomasz Pala <gotar at pld-linux.org>


From jajcus at jajcus.net  Tue Nov 13 20:16:36 2012
From: jajcus at jajcus.net (Jacek Konieczny)
Date: Tue, 13 Nov 2012 20:16:36 +0100
Subject: [kernel] 'perf' utility and other utils from kernel sources
In-Reply-To: <20121113183105.GA24118@polanet.pl>
References: <20121113112501.GC7561@jajo.eggsoft>
 <201211131234.06853.arekm@maven.pl>
 <20121113183105.GA24118@polanet.pl>
Message-ID: <20121113191636.GA2125@lolek.nigdzie>

On Tue, Nov 13, 2012 at 07:31:05PM +0100, Tomasz Pala wrote:
> On Tue, Nov 13, 2012 at 12:34:06 +0100, Arkadiusz Mi?kiewicz wrote:
> 
> >> It seems the 'perf' utility may be a great tool??? but I cannot find it in
> >> PLD. It should be probably built from kernel.spec.
> > 
> > If it's stricly dependant on kernel version then most likely should build from 
> > kernel.spec (and afaik it is dependant).
> 
> It makes use of debug symbols, but it's not _strictly_ dependant. I use
> some older build for quite a long time.

IMHO it still will be convenient to build it from kernel.spec, just not
add strict dependencies to the kernel it was built with.

Greets,
        Jacek


From gotar at polanet.pl  Tue Nov 13 22:57:23 2012
From: gotar at polanet.pl (Tomasz Pala)
Date: Tue, 13 Nov 2012 22:57:23 +0100
Subject: [kernel] 'perf' utility and other utils from kernel sources
In-Reply-To: <20121113191636.GA2125@lolek.nigdzie>
References: <20121113112501.GC7561@jajo.eggsoft>
 <201211131234.06853.arekm@maven.pl>
 <20121113183105.GA24118@polanet.pl>
 <20121113191636.GA2125@lolek.nigdzie>
Message-ID: <20121113215723.GA15750@polanet.pl>

On Tue, Nov 13, 2012 at 20:16:36 +0100, Jacek Konieczny wrote:

>> It makes use of debug symbols, but it's not _strictly_ dependant. I use
>> some older build for quite a long time.
> 
> IMHO it still will be convenient to build it from kernel.spec, just not
> add strict dependencies to the kernel it was built with.

On the other hand, it will make maintaining this subpackage much harder.
Unless there's nothing to maintain here. I wonder if this code would be
extracted from kernel.git in some future, as it's not tightly bound to
the kernel (only it's build-ids). Well, no real difference to me since
using own kernels, consider above only as a bunch of hints.
BTW perf top -e ... is a must-have for complex multi-core setup to hount
for bottlenecks, it's one of the basic tools and this shows how amateur
(or obsoleted) PLD has become... sad.

-- 
Tomasz Pala <gotar at pld-linux.org>


From glen at pld-linux.org  Wed Nov 14 11:11:39 2012
From: glen at pld-linux.org (=?ISO-8859-1?Q?Elan_Ruusam=E4e?=)
Date: Wed, 14 Nov 2012 12:11:39 +0200
Subject: [kernel] 'perf' utility and other utils from kernel sources
In-Reply-To: <20121113215723.GA15750@polanet.pl>
References: <20121113112501.GC7561@jajo.eggsoft>
 <201211131234.06853.arekm@maven.pl> <20121113183105.GA24118@polanet.pl>
 <20121113191636.GA2125@lolek.nigdzie> <20121113215723.GA15750@polanet.pl>
Message-ID: <50A36E5B.2040202@pld-linux.org>

On 13.11.2012 23:57, Tomasz Pala wrote:
> On Tue, Nov 13, 2012 at 20:16:36 +0100, Jacek Konieczny wrote:
>
>>> It makes use of debug symbols, but it's not _strictly_ dependant. I use
>>> some older build for quite a long time.
>> IMHO it still will be convenient to build it from kernel.spec, just not
>> add strict dependencies to the kernel it was built with.
> On the other hand, it will make maintaining this subpackage much harder.
> Unless there's nothing to maintain here. I wonder if this code would be
> extracted from kernel.git in some future, as it's not tightly bound to
> the kernel (only it's build-ids). Well, no real difference to me since
> using own kernels, consider above only as a bunch of hints.
> BTW perf top -e ... is a must-have for complex multi-core setup to hount
> for bottlenecks, it's one of the basic tools and this shows how amateur
> (or obsoleted) PLD has become... sad.
>
i'd prefer also separate spec, monstrous specs with 200 lines of deps is 
definately hard to maintain.
and given linux-libc-headers.spec sample, it's also kernel dependant, 
yet we maintain it in separate package

and i'm sure people who want to maintain kernel don't want to maintain 
tools, which makes the tools packages low quality, no proper deps, etc

-- 
glen



From qboosh at pld-linux.org  Fri Nov 16 21:06:55 2012
From: qboosh at pld-linux.org (Jakub Bogusz)
Date: Fri, 16 Nov 2012 21:06:55 +0100
Subject: rpm5: symlink glob strikes back
Message-ID: <20121116200655.GA14661@stranger.qboosh.pl>

There was a patch before 4.5 -> 5 migration, IIRC dropped as "no longer
needed". But now:

[...]
Processing files: colorhug-client-gui-0.1.13-2.x86_64
error: File not found by glob:
/home/users/qboosh/tmp/colorhug-client-0.1.13-root-qboosh/usr/share/icons/hicolor/*/mimetypes/application-x-ccmx.png

while:

$ ls -l /home/users/qboosh/tmp/colorhug-client-0.1.13-root-qboosh/usr/share/icons/hicolor/*/mimetypes/application-x-ccmx.png
lrwxrwxrwx 1 qboosh users 48 11-16 21:01 /home/users/qboosh/tmp/colorhug-client-0.1.13-root-qboosh/usr/share/icons/hicolor/16x16/mimetypes/application-x-ccmx.png -> /usr/share/icons/hicolor/16x16/apps/colorhug.png
lrwxrwxrwx 1 qboosh users 48 11-16 21:01 /home/users/qboosh/tmp/colorhug-client-0.1.13-root-qboosh/usr/share/icons/hicolor/22x22/mimetypes/application-x-ccmx.png -> /usr/share/icons/hicolor/22x22/apps/colorhug.png
lrwxrwxrwx 1 qboosh users 48 11-16 21:01 /home/users/qboosh/tmp/colorhug-client-0.1.13-root-qboosh/usr/share/icons/hicolor/24x24/mimetypes/application-x-ccmx.png -> /usr/share/icons/hicolor/24x24/apps/colorhug.png
lrwxrwxrwx 1 qboosh users 50 11-16 21:01 /home/users/qboosh/tmp/colorhug-client-0.1.13-root-qboosh/usr/share/icons/hicolor/256x256/mimetypes/application-x-ccmx.png -> /usr/share/icons/hicolor/256x256/apps/colorhug.png
lrwxrwxrwx 1 qboosh users 48 11-16 21:01 /home/users/qboosh/tmp/colorhug-client-0.1.13-root-qboosh/usr/share/icons/hicolor/32x32/mimetypes/application-x-ccmx.png -> /usr/share/icons/hicolor/32x32/apps/colorhug.png
lrwxrwxrwx 1 qboosh users 48 11-16 21:01 /home/users/qboosh/tmp/colorhug-client-0.1.13-root-qboosh/usr/share/icons/hicolor/48x48/mimetypes/application-x-ccmx.png -> /usr/share/icons/hicolor/48x48/apps/colorhug.png
lrwxrwxrwx 1 qboosh users 48 11-16 21:01 /home/users/qboosh/tmp/colorhug-client-0.1.13-root-qboosh/usr/share/icons/hicolor/64x64/mimetypes/application-x-ccmx.png -> /usr/share/icons/hicolor/64x64/apps/colorhug.png

These files don't exist in main system.


-- 
Jakub Bogusz    http://qboosh.pl/


From n3npq at me.com  Fri Nov 16 21:30:22 2012
From: n3npq at me.com (Jeffrey Johnson)
Date: Fri, 16 Nov 2012 15:30:22 -0500
Subject: rpm5: symlink glob strikes back
In-Reply-To: <20121116200655.GA14661@stranger.qboosh.pl>
References: <20121116200655.GA14661@stranger.qboosh.pl>
Message-ID: <D32C7F8F-BAA0-4D7E-8E5A-7FD888C28330@me.com>


On Nov 16, 2012, at 3:06 PM, Jakub Bogusz <qboosh at pld-linux.org> wrote:

> There was a patch before 4.5 -> 5 migration, IIRC dropped as "no longer
> needed". But now:
> 
> [...]
> Processing files: colorhug-client-gui-0.1.13-2.x86_64
> error: File not found by glob:
> /home/users/qboosh/tmp/colorhug-client-0.1.13-root-qboosh/usr/share/icons/hicolor/*/mimetypes/application-x-ccmx.png
> 

The old glibc behavior for glob's and symlinks is
as necessary as always afaik.

Meanwhile, RPM includes the older version of glob in
its libraries for portability with GNU extensions.

Is the internal rpmio glob not good enuf any more?

Or what is the core issue?

73 de Jeff



From arekm at maven.pl  Fri Nov 16 21:40:39 2012
From: arekm at maven.pl (Arkadiusz =?utf-8?q?Mi=C5=9Bkiewicz?=)
Date: Fri, 16 Nov 2012 21:40:39 +0100
Subject: rpm5: symlink glob strikes back
In-Reply-To: <20121116200655.GA14661@stranger.qboosh.pl>
References: <20121116200655.GA14661@stranger.qboosh.pl>
Message-ID: <201211162140.40192.arekm@maven.pl>

On Friday 16 of November 2012, Jakub Bogusz wrote:
> There was a patch before 4.5 -> 5 migration, IIRC dropped as "no longer
> needed". But now:

Looks like rpm-glob.patch got lost.

-- 
Arkadiusz Mi?kiewicz, arekm / maven.pl


From baggins at pld-linux.org  Fri Nov 16 21:43:35 2012
From: baggins at pld-linux.org (Jan =?utf-8?Q?R=C4=99korajski?=)
Date: Fri, 16 Nov 2012 21:43:35 +0100
Subject: rpm5: symlink glob strikes back
In-Reply-To: <201211162140.40192.arekm@maven.pl>
References: <20121116200655.GA14661@stranger.qboosh.pl>
 <201211162140.40192.arekm@maven.pl>
Message-ID: <20121116204334.GA1574@home.lan>

On Fri, 16 Nov 2012, Arkadiusz Mi?kiewicz wrote:

> On Friday 16 of November 2012, Jakub Bogusz wrote:
> > There was a patch before 4.5 -> 5 migration, IIRC dropped as "no longer
> > needed". But now:
> 
> Looks like rpm-glob.patch got lost.

I removed it because rpm5 passed the test-case you gave me.
Looks like that test-case was not hard enough.

-- 
Jan R?korajski                                 | PLD/Linux
SysAdm                                         | http://www.pld-linux.org/
baggins<at>mimuw.edu.pl
baggins<at>pld-linux.org


From n3npq at me.com  Fri Nov 16 22:04:14 2012
From: n3npq at me.com (Jeffrey Johnson)
Date: Fri, 16 Nov 2012 16:04:14 -0500
Subject: rpm5: symlink glob strikes back
In-Reply-To: <20121116204334.GA1574@home.lan>
References: <20121116200655.GA14661@stranger.qboosh.pl>
 <201211162140.40192.arekm@maven.pl> <20121116204334.GA1574@home.lan>
Message-ID: <1BEF7551-A114-4526-8472-6789243162A8@me.com>


On Nov 16, 2012, at 3:43 PM, Jan R?korajski wrote:

> On Fri, 16 Nov 2012, Arkadiusz Mi?kiewicz wrote:
> 
>> On Friday 16 of November 2012, Jakub Bogusz wrote:
>>> There was a patch before 4.5 -> 5 migration, IIRC dropped as "no longer
>>> needed". But now:
>> 
>> Looks like rpm-glob.patch got lost.
> 
> I removed it because rpm5 passed the test-case you gave me.
> Looks like that test-case was not hard enough.
> 

So this whole ancient fuss is about
	++    pglob->gl_lstat = Lstat;
or not? I swear this issue was resolved more than 5y ago.

If the patch were solely to override ->gl_lstat in rpmbuild, then I could
carry the patch. Generally changing the default gl_stat to use Lstat isn't
the right fix imho.

The packaging issue is also easily avoided by using %files -f manifest instead of glob's
with symlinks.

hth

73 de Jeff

> -- 
> Jan R?korajski                                 | PLD/Linux
> SysAdm                                         | http://www.pld-linux.org/
> baggins<at>mimuw.edu.pl
> baggins<at>pld-linux.org
> _______________________________________________
> pld-devel-en mailing list
> pld-devel-en at lists.pld-linux.org
> http://lists.pld-linux.org/mailman/listinfo/pld-devel-en



From glen at pld-linux.org  Sat Nov 17 14:01:27 2012
From: glen at pld-linux.org (=?ISO-8859-1?Q?Elan_Ruusam=E4e?=)
Date: Sat, 17 Nov 2012 15:01:27 +0200
Subject: messed up deps in th
Message-ID: <50A78AA7.8050308@pld-linux.org>

1. tries to uninstall coreutils
rpm-5.4.10-33.x86_64
poldek-0.30-1.rc6.4.x86_64

Installing set #3
Processing dependencies...
fsck-2.21.2-3.x86_64 obsoleted by fsck-2.22.1-1.x86_64
fsck-2.22.1-1.x86_64 marks libblkid-2.22.1-1.x86_64 (cap libblkid = 
2.22.1-1)
  libblkid-2.21.2-3.x86_64 obsoleted by libblkid-2.22.1-1.x86_64
    greedy upgrade libmount-2.21.2-3.x86_64 to 2.22.1-1.x86_64 
(unresolved libblkid = 2.21.2-3)
     libmount-2.21.2-3.x86_64 obsoleted by libmount-2.22.1-1.x86_64
       greedy upgrade mount-2.21.2-3.x86_64 to 2.22.1-1.x86_64 
(unresolved libmount = 2.21.2-3)
        mount-2.21.2-3.x86_64 obsoleted by mount-2.22.1-1.x86_64
    greedy upgrade util-linux-2.21.2-3.x86_64 to 2.22.1-1.x86_64 
(unresolved libblkid = 2.21.2-3)
     util-linux-2.21.2-3.x86_64 obsoleted by util-linux-2.22.1-1.x86_64
     eject-2.1.5-4.x86_64 obsoleted by util-linux-2.22.1-1.x86_64
     coreutils-8.16-1.x86_64 obsoleted by util-linux-2.22.1-1.x86_64
error: /bin/id is required by installed openssh-server-6.0p1-2.x86_64, 
give up
error: fsck-2.22.1-1.x86_64: req libblkid.so.1()(64bit) not found
error: fsck-2.22.1-1.x86_64: req libblkid.so.1(BLKID_1.0)(64bit) not found
error: fsck-2.22.1-1.x86_64: req libblkid.so.1(BLKID_2.17)(64bit) not found
Installing set #4


2. gave error in first place, but resumed with next chunk (different 
machine than #1)
rpm-5.4.10-33.x86_64
poldek-0.30-1.rc6.4.x86_64

Installing set #4
Processing dependencies...
SysVinit-tools-2.88-7.x86_64 obsoleted by SysVinit-tools-2.88-9.x86_64
SysVinit-2.88-7.x86_64 obsoleted by SysVinit-2.88-9.x86_64
coreutils-8.16-1.x86_64 obsoleted by util-linux-2.22.1-1.x86_64
util-linux-2.21.2-3.x86_64 obsoleted by util-linux-2.22.1-1.x86_64
error: fileutils is required by installed 
procps-3.2.8-1.20111124.1.x86_64, give up

Installing set #5
Processing dependencies...
coreutils-8.16-1.x86_64 obsoleted by coreutils-8.19-1.x86_64
coreutils-8.19-1.x86_64 marks util-linux-2.22.1-1.x86_64 (cap util-linux 
 >= 2.22)
  util-linux-2.21.2-3.x86_64 obsoleted by util-linux-2.22.1-1.x86_64
libblkid-2.21.2-3.x86_64 obsoleted by libblkid-2.22.1-1.x86_64
libuuid-2.21.2-3.x86_64 obsoleted by libuuid-2.22.1-1.x86_64
  util-linux-2.22.1-1.x86_64 marks libmount-2.22.1-1.x86_64 (cap 
libmount.so.1()(64bit))
  util-linux-2.22.1-1.x86_64 marks SysVinit-tools-2.88-9.x86_64 (cap 
SysVinit-tools >= 2.88-9)
   SysVinit-tools-2.88-7.x86_64 obsoleted by SysVinit-tools-2.88-9.x86_64
     greedy upgrade SysVinit-2.88-7.x86_64 to 2.88-9.x86_64 (unresolved 
SysVinit-tools = 2.88-7)
      SysVinit-2.88-7.x86_64 obsoleted by SysVinit-2.88-9.x86_64
There are 7 packages to install (4 marked by dependencies), 6 to remove:
I coreutils-8.19-1.x86_64  libblkid-2.22.1-1.x86_64 libuuid-2.22.1-1.x86_64
D SysVinit-2.88-9.x86_64  SysVinit-tools-2.88-9.x86_64 
libmount-2.22.1-1.x86_64  util-linux-2.22.1-1.x86_64
R SysVinit-2.88-7.x86_64  SysVinit-tools-2.88-7.x86_64 
coreutils-8.16-1.x86_64  libblkid-2.21.2-3.x86_64 libuuid-2.21.2-3.x86_64
R util-linux-2.21.2-3.x86_64

3. on same machine as #1

There are 10 packages to install, 11 to remove:
I SysVinit-2.88-9.x86_64  SysVinit-tools-2.88-9.x86_64 
coreutils-8.19-1.x86_64  libblkid-2.22.1-1.x86_64 libmount-2.22.1-1.x86_64
I mount-2.22.1-1.x86_64  openssh-6.1p1-2.x86_64 
openssh-server-6.1p1-2.x86_64  openssh-server-ldap-6.1p1-2.x86_64
I util-linux-2.22.1-1.x86_64
R SysVinit-2.88-7.x86_64  SysVinit-tools-2.88-7.x86_64 
coreutils-8.16-1.x86_64  eject-2.1.5-4.x86_64 libblkid-2.21.2-3.x86_64
R libmount-2.21.2-3.x86_64  mount-2.21.2-3.x86_64 
openssh-6.0p1-2.x86_64  openssh-server-6.0p1-2.x86_64
R openssh-server-ldap-6.0p1-2.x86_64  util-linux-2.21.2-3.x86_64
This operation will use 2.2MB of disk space.
Need to get 6.4MB of archives.
Executing rpm --upgrade -vh --root /...
error: Failed dependencies:
         libblkid = 2.21.2-3 is needed by (installed) fsck-2.21.2-3.x86_64

all operations performed with poldek --upgrade-dist and particle install 
"on"

-- 
glen



From pluto at agmk.net  Sat Nov 17 14:14:27 2012
From: pluto at agmk.net (=?utf-8?B?UGF3ZcWC?= Sikora)
Date: Sat, 17 Nov 2012 14:14:27 +0100
Subject: messed up deps in th
In-Reply-To: <50A78AA7.8050308@pld-linux.org>
References: <50A78AA7.8050308@pld-linux.org>
Message-ID: <2207558.rsn8WFU00Z@localhost>

https://bugs.launchpad.net/pld-linux/+bug/1071000



From glen at pld-linux.org  Sat Nov 17 14:27:31 2012
From: glen at pld-linux.org (=?UTF-8?B?RWxhbiBSdXVzYW3DpGU=?=)
Date: Sat, 17 Nov 2012 15:27:31 +0200
Subject: messed up deps in th
In-Reply-To: <2207558.rsn8WFU00Z@localhost>
References: <50A78AA7.8050308@pld-linux.org> <2207558.rsn8WFU00Z@localhost>
Message-ID: <50A790C3.8010105@pld-linux.org>

On 11/17/2012 03:14 PM, Pawe? Sikora wrote:
> https://bugs.launchpad.net/pld-linux/+bug/1071000
>

well, imho first case can be easily fixed by removing unversioned provides: coreutils-su



-- 
glen



From baggins at pld-linux.org  Tue Nov 27 14:16:10 2012
From: baggins at pld-linux.org (Jan =?utf-8?Q?R=C4=99korajski?=)
Date: Tue, 27 Nov 2012 14:16:10 +0100
Subject: rpm5 Berkeley DB downgrade
Message-ID: <20121127131610.GM11369@sith.mimuw.edu.pl>

Hi,
Due to incompatibilities between BDB 5.3 and rpm5 and problems they cause
(DB_BUFFER_SMALL errors causing sometimes partial upgrades, ntpd package
causing rpm5 loop on upgrade) I am forced to downgrade rpm's BDB to 5.2
which is well tested and interoperating properly.

In the next few days I will prepare packages with rpm, poldek and vserver-utils
that will allow clean downgrade (and of course upgrade from rpm4).

When you see rpm and poldek with fractional releases (35.X and rc6.4.Y) in
th-test, those will be using BDB 5.2, so please test them and report
if anything wrong happens.

-- 
Jan R?korajski                                 | PLD/Linux
SysAdm                                         | http://www.pld-linux.org/
baggins<at>mimuw.edu.pl
baggins<at>pld-linux.org


From pawel at viop.pl  Tue Nov 27 15:00:02 2012
From: pawel at viop.pl (=?UTF-8?B?UGF3ZcWCIEtvxZtrYQ==?=)
Date: Tue, 27 Nov 2012 15:00:02 +0100
Subject: rpm5 Berkeley DB downgrade
In-Reply-To: <20121127131610.GM11369@sith.mimuw.edu.pl>
References: <20121127131610.GM11369@sith.mimuw.edu.pl>
Message-ID: <CAP53uYURGTpJ2ZvzvTi4DBq8s1aTZ80Hky33CrGc=KDMxqnw_w@mail.gmail.com>

A czy naprawi? si? b??dy przy budowaniu pakiet?w?
U mnie w aktualnym th wewn?trz vservera wszystkie pr?by zbudowania pakiet?w
ko?cz? si? tak:

Executing(%doc): /bin/sh -e /tmp/rpm-tmp.19135
+ umask 022
+ cd /home/users/builder/rpm/BUILD
+ cd mtr-0.82
+ DOCDIR=/tmp/mtr-0.82-root-builder/usr/share/doc/mtr-0.82
+ export DOCDIR
+ /bin/mkdir -p /tmp/mtr-0.82-root-builder/usr/share/doc/mtr-0.82
+ /usr/lib/rpm/compress-doc
Compressing documentation in
/tmp/mtr-0.82-root-builder/usr/share/doc/mtr-0.82...

Documentation compressed.
+ cp -pr AUTHORS NEWS README SECURITY
/tmp/mtr-0.82-root-builder/usr/share/doc/mtr-0.82
+ exit 0
skipping /usr/share/doc/mtr-0.82 requires detection
Provides: elf(buildid) = bb48319bc883dee191b3dd28397e402b41dba6c8
Requires: libc.so.6 libc.so.6(GLIBC_2.0) libc.so.6(GLIBC_2.1)
libc.so.6(GLIBC_2.15) libc.so.6(GLIBC_2.2) libc.so.6(GLIBC_2.3)
libc.so.6(GLIBC_2.3.4) libc.so.6(GLIBC_2.4) libm.so.6 libm.so.6(GLIBC_2.0)
libncurses.so.5 libresolv.so.2 libresolv.so.2(GLIBC_2.2) rtld(GNU_HASH)
Obsoletes: mtr-ncurses
Processing files: mtr-debuginfo-0.82-2.i686
Checking for unpackaged file(s): /usr/lib/rpm/check-files
/tmp/mtr-0.82-root-builder
rpmbuild: ./rpmio_internal.h:307: fdGetFp: Assertion `fd != ((void *)0) &&
fd->magic == 0x04463138' failed.
Aborted


3x tworzy?em vserver na 2 maszynach i zawsze mam tak samo ;)


2012/11/27 Jan R?korajski <baggins at pld-linux.org>

> Hi,
> Due to incompatibilities between BDB 5.3 and rpm5 and problems they cause
> (DB_BUFFER_SMALL errors causing sometimes partial upgrades, ntpd package
> causing rpm5 loop on upgrade) I am forced to downgrade rpm's BDB to 5.2
> which is well tested and interoperating properly.
>
> In the next few days I will prepare packages with rpm, poldek and
> vserver-utils
> that will allow clean downgrade (and of course upgrade from rpm4).
>
> When you see rpm and poldek with fractional releases (35.X and rc6.4.Y) in
> th-test, those will be using BDB 5.2, so please test them and report
> if anything wrong happens.
>
> --
> Jan R?korajski                                 | PLD/Linux
> SysAdm                                         | http://www.pld-linux.org/
> baggins<at>mimuw.edu.pl
> baggins<at>pld-linux.org
> _______________________________________________
> pld-devel-pl mailing list
> pld-devel-pl at lists.pld-linux.org
> http://lists.pld-linux.org/mailman/listinfo/pld-devel-pl
>


From n3npq at me.com  Tue Nov 27 16:56:14 2012
From: n3npq at me.com (Jeffrey Johnson)
Date: Tue, 27 Nov 2012 10:56:14 -0500
Subject: rpm5 Berkeley DB downgrade
In-Reply-To: <CAP53uYURGTpJ2ZvzvTi4DBq8s1aTZ80Hky33CrGc=KDMxqnw_w@mail.gmail.com>
References: <20121127131610.GM11369@sith.mimuw.edu.pl>
 <CAP53uYURGTpJ2ZvzvTi4DBq8s1aTZ80Hky33CrGc=KDMxqnw_w@mail.gmail.com>
Message-ID: <98D81F31-89D1-4DC9-806E-E043104F9476@me.com>

English please.

On Nov 27, 2012, at 9:00 AM, Pawe? Ko?ka wrote:

> A czy naprawi? si? b??dy przy budowaniu pakiet?w?
> U mnie w aktualnym th wewn?trz vservera wszystkie pr?by zbudowania pakiet?w
> ko?cz? si? tak:
> 
> Executing(%doc): /bin/sh -e /tmp/rpm-tmp.19135
> + umask 022
> + cd /home/users/builder/rpm/BUILD
> + cd mtr-0.82
> + DOCDIR=/tmp/mtr-0.82-root-builder/usr/share/doc/mtr-0.82
> + export DOCDIR
> + /bin/mkdir -p /tmp/mtr-0.82-root-builder/usr/share/doc/mtr-0.82
> + /usr/lib/rpm/compress-doc
> Compressing documentation in
> /tmp/mtr-0.82-root-builder/usr/share/doc/mtr-0.82...
> 
> Documentation compressed.
> + cp -pr AUTHORS NEWS README SECURITY
> /tmp/mtr-0.82-root-builder/usr/share/doc/mtr-0.82
> + exit 0
> skipping /usr/share/doc/mtr-0.82 requires detection
> Provides: elf(buildid) = bb48319bc883dee191b3dd28397e402b41dba6c8
> Requires: libc.so.6 libc.so.6(GLIBC_2.0) libc.so.6(GLIBC_2.1)
> libc.so.6(GLIBC_2.15) libc.so.6(GLIBC_2.2) libc.so.6(GLIBC_2.3)
> libc.so.6(GLIBC_2.3.4) libc.so.6(GLIBC_2.4) libm.so.6 libm.so.6(GLIBC_2.0)
> libncurses.so.5 libresolv.so.2 libresolv.so.2(GLIBC_2.2) rtld(GNU_HASH)
> Obsoletes: mtr-ncurses
> Processing files: mtr-debuginfo-0.82-2.i686
> Checking for unpackaged file(s): /usr/lib/rpm/check-files
> /tmp/mtr-0.82-root-builder
> rpmbuild: ./rpmio_internal.h:307: fdGetFp: Assertion `fd != ((void *)0) &&
> fd->magic == 0x04463138' failed.
> Aborted
> 

This is an assertion failure that shows up when a FD_t I/O handle is
closed more than once, which can/does happen on obscure error pathways.

73 de Jeff
> 
> 3x tworzy?em vserver na 2 maszynach i zawsze mam tak samo ;)
> 
> 
> 2012/11/27 Jan R?korajski <baggins at pld-linux.org>
> 
>> Hi,
>> Due to incompatibilities between BDB 5.3 and rpm5 and problems they cause
>> (DB_BUFFER_SMALL errors causing sometimes partial upgrades, ntpd package
>> causing rpm5 loop on upgrade) I am forced to downgrade rpm's BDB to 5.2
>> which is well tested and interoperating properly.
>> 
>> In the next few days I will prepare packages with rpm, poldek and
>> vserver-utils
>> that will allow clean downgrade (and of course upgrade from rpm4).
>> 
>> When you see rpm and poldek with fractional releases (35.X and rc6.4.Y) in
>> th-test, those will be using BDB 5.2, so please test them and report
>> if anything wrong happens.
>> 
>> --
>> Jan R?korajski                                 | PLD/Linux
>> SysAdm                                         | http://www.pld-linux.org/
>> baggins<at>mimuw.edu.pl
>> baggins<at>pld-linux.org
>> _______________________________________________
>> pld-devel-pl mailing list
>> pld-devel-pl at lists.pld-linux.org
>> http://lists.pld-linux.org/mailman/listinfo/pld-devel-pl
>> 
> _______________________________________________
> pld-devel-en mailing list
> pld-devel-en at lists.pld-linux.org
> http://lists.pld-linux.org/mailman/listinfo/pld-devel-en



From n3npq at me.com  Tue Nov 27 16:59:50 2012
From: n3npq at me.com (Jeffrey Johnson)
Date: Tue, 27 Nov 2012 10:59:50 -0500
Subject: rpm5 Berkeley DB downgrade
In-Reply-To: <20121127131610.GM11369@sith.mimuw.edu.pl>
References: <20121127131610.GM11369@sith.mimuw.edu.pl>
Message-ID: <3B70723B-77E4-4A15-BCD2-11D77CE27EE9@me.com>


On Nov 27, 2012, at 8:16 AM, Jan R?korajski wrote:

> Hi,
> Due to incompatibilities between BDB 5.3 and rpm5 and problems they cause
> (DB_BUFFER_SMALL errors causing sometimes partial upgrades, ntpd package
> causing rpm5 loop on upgrade) I am forced to downgrade rpm's BDB to 5.2
> which is well tested and interoperating properly.

This is incompatibilities between db-5.2 and db-5.3, not between rpm and bdb.

Reverting isn't the only solution (as you know), but is likely the best/easiest solution.

Meanwhile the problem (in db-5.3.x) won't ever be fixed until I get a reproducer on
my devel box.

73 de Jeff

> 
> In the next few days I will prepare packages with rpm, poldek and vserver-utils
> that will allow clean downgrade (and of course upgrade from rpm4).
> 
> When you see rpm and poldek with fractional releases (35.X and rc6.4.Y) in
> th-test, those will be using BDB 5.2, so please test them and report
> if anything wrong happens.
> 
> -- 
> Jan R?korajski                                 | PLD/Linux
> SysAdm                                         | http://www.pld-linux.org/
> baggins<at>mimuw.edu.pl
> baggins<at>pld-linux.org
> _______________________________________________
> pld-devel-en mailing list
> pld-devel-en at lists.pld-linux.org
> http://lists.pld-linux.org/mailman/listinfo/pld-devel-en



From glen at pld-linux.org  Tue Nov 27 17:17:27 2012
From: glen at pld-linux.org (=?UTF-8?B?RWxhbiBSdXVzYW3DpGU=?=)
Date: Tue, 27 Nov 2012 18:17:27 +0200
Subject: rpm5 Berkeley DB downgrade
In-Reply-To: <3B70723B-77E4-4A15-BCD2-11D77CE27EE9@me.com>
References: <20121127131610.GM11369@sith.mimuw.edu.pl>
 <3B70723B-77E4-4A15-BCD2-11D77CE27EE9@me.com>
Message-ID: <50B4E797.7040901@pld-linux.org>

On 27.11.2012 17:59, Jeffrey Johnson wrote:
> Meanwhile the problem (in db-5.3.x) won't ever be fixed until I get a reproducer on
> my devel box.
i gave you all the bits you asked and need. why do don't you utilize 
them? i haven't heard back from you either or that you have any kind of 
problems

-- 
glen



From n3npq at me.com  Tue Nov 27 17:31:22 2012
From: n3npq at me.com (Jeffrey Johnson)
Date: Tue, 27 Nov 2012 11:31:22 -0500
Subject: rpm5 Berkeley DB downgrade
In-Reply-To: <50B4E797.7040901@pld-linux.org>
References: <20121127131610.GM11369@sith.mimuw.edu.pl>
 <3B70723B-77E4-4A15-BCD2-11D77CE27EE9@me.com> <50B4E797.7040901@pld-linux.org>
Message-ID: <A4932031-F73F-46CC-B258-1469C520138E@me.com>


On Nov 27, 2012, at 11:17 AM, Elan Ruusam?e wrote:

> On 27.11.2012 17:59, Jeffrey Johnson wrote:
>> Meanwhile the problem (in db-5.3.x) won't ever be fixed until I get a reproducer on
>> my devel box.
> i gave you all the bits you asked and need. why do don't you utilize them? i haven't heard back from you either or that you have any kind of problems
> 

Yes you did. I haven't used because other priorities (like getting paid) intrude.

73 de Jeff

From glen at pld-linux.org  Wed Nov 28 17:01:46 2012
From: glen at pld-linux.org (=?ISO-8859-1?Q?Elan_Ruusam=E4e?=)
Date: Wed, 28 Nov 2012 18:01:46 +0200
Subject: rpm5 triggers
Message-ID: <50B6356A.202@pld-linux.org>

where's the doc describing how triggers in rpm5 work? them seem to 
misbehave, ie $1 should be 1 on package first install, but my test shows 
that it is 0 with rpm5

rpm-5.4.10-35.x86_64


-- 
glen



From glen at pld-linux.org  Wed Nov 28 17:09:47 2012
From: glen at pld-linux.org (=?ISO-8859-1?Q?Elan_Ruusam=E4e?=)
Date: Wed, 28 Nov 2012 18:09:47 +0200
Subject: rpm5 triggers
In-Reply-To: <50B6356A.202@pld-linux.org>
References: <50B6356A.202@pld-linux.org>
Message-ID: <50B6374B.1070602@pld-linux.org>

On 28.11.2012 18:01, Elan Ruusam?e wrote:
> where's the doc describing how triggers in rpm5 work? them seem to 
> misbehave, ie $1 should be 1 on package first install, but my test 
> shows that it is 0 with rpm5
>
> rpm-5.4.10-35.x86_64
>
>
$1 which should be 1 on first install, is 0 with rpm5, which is breaking 
our webapps install trigger

webapps macros defined here:
http://git.pld-linux.org/?p=packages/rpm-build-macros.git;a=blob;f=rpm.macros;h=5a9c6d2c1e507fb8460c212b79a8f2083c3f8d78;hb=0a9bb9747529d8f89a74f6db5f317abe1454d8ac#l1244

test.spec from 
http://git.pld-linux.org/?p=packages/test.git;a=shortlog;h=refs/heads/rpm5-trigger

resulting rpm:
http://carme.pld-linux.org/~glen/test-0.1-0.1.noarch.rpm

installing with rpm-0:4.5-70.amd64:

$ sudo rpm -Uhv test-0.1-0.1.noarch.rpm
warning: test-0.1-0.1.noarch.rpm: Header V4 DSA signature: NOKEY, key ID 
0e26e127
Preparing... ########################################### [100%]
    1:test ########################################### [100%]
+ echo triggerin:%{N}-%{V}-%{R}
triggerin:%{N}-%{V}-%{R}
+ [ -n  ]
+ [ 1 = 1 ]
+ [ 1 = 1 ]
+ /usr/sbin/webapp register lighttpd test
+ [ -L /etc/lighttpd/webapps.d/test.conf ]
+ [ 1 -lt 2 ]
+ skip_auto_restart
+ [ no = no ]
+ [ ! -x /bin/systemd_booted ]
+ [ -f /var/lock/subsys/lighttpd ]
+ /sbin/service lighttpd reload
+ >&2
Reloading Lighttpd Web Server service..............................[ DIED ]
...

installed with rpm-5.4.10-35.x86_64 :
# rpm -Uhv test-0.1-0.1.noarch.rpm
Preparing... ########################################### [100%]
    1:test ########################################### [100%]
+ echo triggerin:%{N}-%{V}-%{R}
triggerin:%{N}-%{V}-%{R}
+ [ -n  ]
+ [ 0 = 1 ]
+ [ -L /etc/lighttpd/webapps.d/test.conf ]
...

-- 
glen



From n3npq at me.com  Wed Nov 28 17:14:07 2012
From: n3npq at me.com (Jeffrey Johnson)
Date: Wed, 28 Nov 2012 11:14:07 -0500
Subject: rpm5 triggers
In-Reply-To: <50B6356A.202@pld-linux.org>
References: <50B6356A.202@pld-linux.org>
Message-ID: <41117BC5-50CC-4149-8828-9B96AD21DDB8@me.com>


On Nov 28, 2012, at 11:01 AM, Elan Ruusam?e <glen at pld-linux.org> wrote:

> where's the doc describing how triggers in rpm5 work? them seem to misbehave, ie $1 should be 1 on package first install, but my test shows that it is 0 with rpm5
> 


The doc is right where its always been: doc/triggers.

You need to describe your test. Trigger behavior is
explicitly tested before every release at rpm5.org.

73 de Jeff



From glen at pld-linux.org  Wed Nov 28 17:18:56 2012
From: glen at pld-linux.org (=?ISO-8859-1?Q?Elan_Ruusam=E4e?=)
Date: Wed, 28 Nov 2012 18:18:56 +0200
Subject: rpm5 triggers
In-Reply-To: <41117BC5-50CC-4149-8828-9B96AD21DDB8@me.com>
References: <50B6356A.202@pld-linux.org>
 <41117BC5-50CC-4149-8828-9B96AD21DDB8@me.com>
Message-ID: <50B63970.9020603@pld-linux.org>

On 28.11.2012 18:14, Jeffrey Johnson wrote:
> On Nov 28, 2012, at 11:01 AM, Elan Ruusam?e <glen at pld-linux.org> wrote:
>
>> where's the doc describing how triggers in rpm5 work? them seem to misbehave, ie $1 should be 1 on package first install, but my test shows that it is 0 with rpm5
>>
>
> The doc is right where its always been: doc/triggers.

good, at least i'm not stupid then, expecting things to work and not 
reading proper docs :)

for the record, no functional change in that file:
$ diff -u triggers-rpm?
--- triggers-rpm4    2007-05-25 20:35:58.000000000 +0300
+++ triggers-rpm5    2008-02-18 02:40:45.000000000 +0200
@@ -105,11 +105,11 @@
  Trigger specifications are of the form:

  \verbatim
-    %trigger{un|in|postun} [[-n] <subpackage>] [-p <program>] -- <trigger>
+    %trigger{un|in|postun} [[-n] SUBPKGSUFFIX] [-p PROGRAM] -- TRIGGER
  \endverbatim

  The -n and -p arguments are the same as for %post scripts.  The
-<trigger> portion is syntactically equivalent to a "Requires"
+TRIGGER portion is syntactically equivalent to a "Requires"
  specification (version numbers may be used). If multiple items are
  given (comma separated), the trigger is run when *any* of those
  conditions becomes true (the , can be read as "or"). For example:
(END)


> You need to describe your test. Trigger behavior is
> explicitly tested before every release at rpm5.org.
yeah. it could be pld patches as well (baggins applied bunch of "useful 
ones" from mandriva repository.


-- 
glen



From n3npq at me.com  Wed Nov 28 17:26:31 2012
From: n3npq at me.com (Jeffrey Johnson)
Date: Wed, 28 Nov 2012 11:26:31 -0500
Subject: rpm5 triggers
In-Reply-To: <50B63970.9020603@pld-linux.org>
References: <50B6356A.202@pld-linux.org>
 <41117BC5-50CC-4149-8828-9B96AD21DDB8@me.com> <50B63970.9020603@pld-linux.org>
Message-ID: <BC51BF88-C51C-4E43-A480-781996B44112@me.com>


On Nov 28, 2012, at 11:18 AM, Elan Ruusam?e <glen at pld-linux.org> wrote:

> On 28.11.2012 18:14, Jeffrey Johnson wrote:
>> On Nov 28, 2012, at 11:01 AM, Elan Ruusam?e <glen at pld-linux.org> wrote:
>> 
>>> where's the doc describing how triggers in rpm5 work? them seem to misbehave, ie $1 should be 1 on package first install, but my test shows that it is 0 with rpm5
>>> 
>> 
>> The doc is right where its always been: doc/triggers.
> 
> good, at least i'm not stupid then, expecting things to work and not reading proper docs :)
> 
> for the record, no functional change in that file:
> $ diff -u triggers-rpm?
> --- triggers-rpm4    2007-05-25 20:35:58.000000000 +0300
> +++ triggers-rpm5    2008-02-18 02:40:45.000000000 +0200
> @@ -105,11 +105,11 @@
> Trigger specifications are of the form:
> 
> \verbatim
> -    %trigger{un|in|postun} [[-n] <subpackage>] [-p <program>] -- <trigger>
> +    %trigger{un|in|postun} [[-n] SUBPKGSUFFIX] [-p PROGRAM] -- TRIGGER
> \endverbatim
> 
> The -n and -p arguments are the same as for %post scripts.  The
> -<trigger> portion is syntactically equivalent to a "Requires"
> +TRIGGER portion is syntactically equivalent to a "Requires"
> specification (version numbers may be used). If multiple items are
> given (comma separated), the trigger is run when *any* of those
> conditions becomes true (the , can be read as "or"). For example:
> (END)
> 

And the doco is just as deficient as ever because:

1) there is a 4th trigger called %triggerprein

2) triggers are fired by provides, not package names

3) previously (like >5y ago) only a single trigger was run
when multiple trigger relations between 2 packages existed;
all triggers are now run.

4) triggers on directory/file paths (and even globs) are implemented


> 
>> You need to describe your test. Trigger behavior is
>> explicitly tested before every release at rpm5.org.
> yeah. it could be pld patches as well (baggins applied bunch of "useful ones" from mandriva repository.
> 

Could be a lot of things
> 
> -- 
> glen
> 
> _______________________________________________
> pld-devel-en mailing list
> pld-devel-en at lists.pld-linux.org
> http://lists.pld-linux.org/mailman/listinfo/pld-devel-en



From baggins at pld-linux.org  Wed Nov 28 18:16:17 2012
From: baggins at pld-linux.org (Jan =?utf-8?Q?R=C4=99korajski?=)
Date: Wed, 28 Nov 2012 18:16:17 +0100
Subject: rpm5 triggers
In-Reply-To: <50B63970.9020603@pld-linux.org>
References: <50B6356A.202@pld-linux.org>
 <41117BC5-50CC-4149-8828-9B96AD21DDB8@me.com>
 <50B63970.9020603@pld-linux.org>
Message-ID: <20121128171617.GA1540@home.lan>

On Wed, 28 Nov 2012, Elan Ruusam?e wrote:

> On 28.11.2012 18:14, Jeffrey Johnson wrote:
> > On Nov 28, 2012, at 11:01 AM, Elan Ruusam?e <glen at pld-linux.org> wrote:
> >
> >> where's the doc describing how triggers in rpm5 work? them seem to misbehave, ie $1 should be 1 on package first install, but my test shows that it is 0 with rpm5
> >>
> >
> > The doc is right where its always been: doc/triggers.
> 
> good, at least i'm not stupid then, expecting things to work and not 
> reading proper docs :)
> 
> for the record, no functional change in that file:
> $ diff -u triggers-rpm?
> --- triggers-rpm4    2007-05-25 20:35:58.000000000 +0300
> +++ triggers-rpm5    2008-02-18 02:40:45.000000000 +0200
> @@ -105,11 +105,11 @@
>   Trigger specifications are of the form:
> 
>   \verbatim
> -    %trigger{un|in|postun} [[-n] <subpackage>] [-p <program>] -- <trigger>
> +    %trigger{un|in|postun} [[-n] SUBPKGSUFFIX] [-p PROGRAM] -- TRIGGER
>   \endverbatim
> 
>   The -n and -p arguments are the same as for %post scripts.  The
> -<trigger> portion is syntactically equivalent to a "Requires"
> +TRIGGER portion is syntactically equivalent to a "Requires"
>   specification (version numbers may be used). If multiple items are
>   given (comma separated), the trigger is run when *any* of those
>   conditions becomes true (the , can be read as "or"). For example:
> (END)
> 
> 
> > You need to describe your test. Trigger behavior is
> > explicitly tested before every release at rpm5.org.
> yeah. it could be pld patches as well (baggins applied bunch of "useful 
> ones" from mandriva repository.

Nothing in those patches touch execution of triggers, so it must be
something in rpm5 itself that changed.

-- 
Jan R?korajski                                 | PLD/Linux
SysAdm                                         | http://www.pld-linux.org/
baggins<at>mimuw.edu.pl
baggins<at>pld-linux.org


From n3npq at me.com  Wed Nov 28 18:19:17 2012
From: n3npq at me.com (Jeffrey Johnson)
Date: Wed, 28 Nov 2012 12:19:17 -0500
Subject: rpm5 triggers
In-Reply-To: <20121128171617.GA1540@home.lan>
References: <50B6356A.202@pld-linux.org>
 <41117BC5-50CC-4149-8828-9B96AD21DDB8@me.com> <50B63970.9020603@pld-linux.org>
 <20121128171617.GA1540@home.lan>
Message-ID: <8BB9A282-1668-4EBB-A57D-85C589091909@me.com>


On Nov 28, 2012, at 12:16 PM, Jan R?korajski <baggins at pld-linux.org> wrote:

> 
> Nothing in those patches touch execution of triggers, so it must be
> something in rpm5 itself that changed.

The diagnostic reasoning is faulty because it doesn't lead anywhere.

I have yet to hear what the test is/was.

73 de Jeff



From baggins at pld-linux.org  Wed Nov 28 18:53:23 2012
From: baggins at pld-linux.org (Jan =?utf-8?Q?R=C4=99korajski?=)
Date: Wed, 28 Nov 2012 18:53:23 +0100
Subject: rpm5 triggers
In-Reply-To: <8BB9A282-1668-4EBB-A57D-85C589091909@me.com>
References: <50B6356A.202@pld-linux.org>
 <41117BC5-50CC-4149-8828-9B96AD21DDB8@me.com>
 <50B63970.9020603@pld-linux.org> <20121128171617.GA1540@home.lan>
 <8BB9A282-1668-4EBB-A57D-85C589091909@me.com>
Message-ID: <20121128175323.GB1540@home.lan>

On Wed, 28 Nov 2012, Jeffrey Johnson wrote:

> 
> On Nov 28, 2012, at 12:16 PM, Jan R?korajski <baggins at pld-linux.org> wrote:
> 
> > 
> > Nothing in those patches touch execution of triggers, so it must be
> > something in rpm5 itself that changed.
> 
> The diagnostic reasoning is faulty because it doesn't lead anywhere.
> 
> I have yet to hear what the test is/was.

Test is simple:

test.spec contains:

%triggerin -- lighttpd
set -x
echo "#: $#"
echo "1: $1"
echo "2: $2"

Now if lighttpd is already installed then when installing test package
$1 and $2 should be 1, but $1 is 0 in rpm5 (contrary to documentation).
Same if test is already installed and installing lighttpd package, this
time $2 is 0 in rpm5.

Looks like those args are calculated incorrectly.

-- 
Jan R?korajski                                 | PLD/Linux
SysAdm                                         | http://www.pld-linux.org/
baggins<at>mimuw.edu.pl
baggins<at>pld-linux.org


From n3npq at me.com  Wed Nov 28 18:59:21 2012
From: n3npq at me.com (Jeffrey Johnson)
Date: Wed, 28 Nov 2012 12:59:21 -0500
Subject: rpm5 triggers
In-Reply-To: <20121128175323.GB1540@home.lan>
References: <50B6356A.202@pld-linux.org>
 <41117BC5-50CC-4149-8828-9B96AD21DDB8@me.com> <50B63970.9020603@pld-linux.org>
 <20121128171617.GA1540@home.lan> <8BB9A282-1668-4EBB-A57D-85C589091909@me.com>
 <20121128175323.GB1540@home.lan>
Message-ID: <A2EBE76E-211E-446A-921B-941FC73EDC00@me.com>


On Nov 28, 2012, at 12:53 PM, Jan R?korajski <baggins at pld-linux.org> wrote:

> On Wed, 28 Nov 2012, Jeffrey Johnson wrote:
> 
>> 
>> On Nov 28, 2012, at 12:16 PM, Jan R?korajski <baggins at pld-linux.org> wrote:
>> 
>>> 
>>> Nothing in those patches touch execution of triggers, so it must be
>>> something in rpm5 itself that changed.
>> 
>> The diagnostic reasoning is faulty because it doesn't lead anywhere.
>> 
>> I have yet to hear what the test is/was.
> 
> Test is simple:
> 
> test.spec contains:
> 
> %triggerin -- lighttpd
> set -x
> echo "#: $#"
> echo "1: $1"
> echo "2: $2"
> 
> Now if lighttpd is already installed then when installing test package
> $1 and $2 should be 1, but $1 is 0 in rpm5 (contrary to documentation).
> Same if test is already installed and installing lighttpd package, this
> time $2 is 0 in rpm5.
> 
> Looks like those args are calculated incorrectly.

The test is _NOT_ simple because triggers fire depending
on what is installed and present in an rpmdb.

There are also failure modes when the retrieval, not
the calculation, are incorrect.

Again the starting point is -vv output, and then verifying
the retrievals are actually correct by adding "debug" to the
Triggername index:
	%_dbi_config_3_Triggername %{_dbi_btconfig} %{?_bt_dupsort}

Premature guessing isn't going to solve any RPM problems. And I
have no idea what your rpmdb contains.

73 de Jeff

From baggins at pld-linux.org  Wed Nov 28 19:51:54 2012
From: baggins at pld-linux.org (Jan =?utf-8?Q?R=C4=99korajski?=)
Date: Wed, 28 Nov 2012 19:51:54 +0100
Subject: rpm5 triggers
In-Reply-To: <A2EBE76E-211E-446A-921B-941FC73EDC00@me.com>
References: <50B6356A.202@pld-linux.org>
 <41117BC5-50CC-4149-8828-9B96AD21DDB8@me.com>
 <50B63970.9020603@pld-linux.org> <20121128171617.GA1540@home.lan>
 <8BB9A282-1668-4EBB-A57D-85C589091909@me.com>
 <20121128175323.GB1540@home.lan>
 <A2EBE76E-211E-446A-921B-941FC73EDC00@me.com>
Message-ID: <20121128185154.GD1540@home.lan>

On Wed, 28 Nov 2012, Jeffrey Johnson wrote:

> 
> On Nov 28, 2012, at 12:53 PM, Jan R?korajski <baggins at pld-linux.org> wrote:
> 
> > On Wed, 28 Nov 2012, Jeffrey Johnson wrote:
> > 
> >> 
> >> On Nov 28, 2012, at 12:16 PM, Jan R?korajski <baggins at pld-linux.org> wrote:
> >> 
> >>> 
> >>> Nothing in those patches touch execution of triggers, so it must be
> >>> something in rpm5 itself that changed.
> >> 
> >> The diagnostic reasoning is faulty because it doesn't lead anywhere.
> >> 
> >> I have yet to hear what the test is/was.
> > 
> > Test is simple:
> > 
> > test.spec contains:
> > 
> > %triggerin -- lighttpd
> > set -x
> > echo "#: $#"
> > echo "1: $1"
> > echo "2: $2"
> > 
> > Now if lighttpd is already installed then when installing test package
> > $1 and $2 should be 1, but $1 is 0 in rpm5 (contrary to documentation).
> > Same if test is already installed and installing lighttpd package, this
> > time $2 is 0 in rpm5.
> > 
> > Looks like those args are calculated incorrectly.
> 
> The test is _NOT_ simple because triggers fire depending
> on what is installed and present in an rpmdb.
> 
> There are also failure modes when the retrieval, not
> the calculation, are incorrect.
> 
> Again the starting point is -vv output, and then verifying
> the retrievals are actually correct by adding "debug" to the
> Triggername index:
> 	%_dbi_config_3_Triggername %{_dbi_btconfig} %{?_bt_dupsort}
> 
> Premature guessing isn't going to solve any RPM problems. And I
> have no idea what your rpmdb contains.
> 

Ok, so here come attachments (gzipped due to list limits):
- test.spec with triggers
- lighttpd-install, -vvv output of installing target package (lighttpd)
  when source package (test) is already installed
- lighttpd-uninstall, -vvv output of removing target package (lighttpd)
  when source package (test) is already installed
- test-install, -vvv output of installing source package (test) when
  target package (lighttpd) is already installed
- test-uninstall, -vvv output of removing source package (test) when
  target package (lighttpd) is already installed

Numbers passed as arguments to triggers in test.spec are incorrect
according to docs.

-- 
Jan R?korajski                                 | PLD/Linux
SysAdm                                         | http://www.pld-linux.org/
baggins<at>mimuw.edu.pl
baggins<at>pld-linux.org
-------------- next part --------------
Summary:	webapp test
Name:		test
Version:	0.1
Release:	0.1
License:	GPL
Group:		Applications/System
BuildRequires:	rpmbuild(macros) >= 1.268
Requires:	webapps
BuildArch:	noarch
BuildRoot:	%{tmpdir}/%{name}-%{version}-root-%(id -u -n)

%define		_webapps	/etc/webapps
%define		_webapp		%{name}
%define		_sysconfdir	%{_webapps}/%{_webapp}

%description
zz

%prep
%setup -qcT
touch lighttpd.conf

%install
rm -rf $RPM_BUILD_ROOT
install -d $RPM_BUILD_ROOT%{_sysconfdir}
cp -p lighttpd.conf $RPM_BUILD_ROOT%{_sysconfdir}/lighttpd.conf

%triggerin -- lighttpd
set -x
echo "#: $#"
echo "1: $1"
echo "2: $2"
echo triggerin:%{name}-%{version}-%{release}

%triggerun -- lighttpd
set -x
echo "#: $#"
echo "1: $1"
echo "2: $2"
echo triggerun:%{name}-%{version}-%{release}

%clean
rm -rf $RPM_BUILD_ROOT

%files
%defattr(644,root,root,755)
%dir %attr(750,root,http) %{_sysconfdir}
%attr(640,root,root) %config(noreplace) %verify(not md5 mtime size) %{_sysconfdir}/lighttpd.conf
-------------- next part --------------
A non-text attachment was scrubbed...
Name: test-install.gz
Type: application/x-gzip
Size: 2277 bytes
Desc: not available
URL: </mailman/pipermail/pld-devel-en/attachments/20121128/eafaddde/attachment.bin>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: test-uninstall.gz
Type: application/x-gzip
Size: 1900 bytes
Desc: not available
URL: </mailman/pipermail/pld-devel-en/attachments/20121128/eafaddde/attachment-0001.bin>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: lighttpd-install.gz
Type: application/x-gzip
Size: 6553 bytes
Desc: not available
URL: </mailman/pipermail/pld-devel-en/attachments/20121128/eafaddde/attachment-0002.bin>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: lighttpd-uninstall.gz
Type: application/x-gzip
Size: 3770 bytes
Desc: not available
URL: </mailman/pipermail/pld-devel-en/attachments/20121128/eafaddde/attachment-0003.bin>

From baggins at pld-linux.org  Fri Nov 30 13:41:53 2012
From: baggins at pld-linux.org (Jan =?utf-8?Q?R=C4=99korajski?=)
Date: Fri, 30 Nov 2012 13:41:53 +0100
Subject: rpm-5.4.10-22.i686 loops forever when installing
 ntpd-4.2.6p5-5.i686.rpm
In-Reply-To: <50846358.8090607@pld-linux.org>
References: <50846358.8090607@pld-linux.org>
Message-ID: <20121130124153.GG29634@sith.mimuw.edu.pl>

On Mon, 22 Oct 2012, Elan Ruusam?e wrote:

> Script started on Sun 21 Oct 2012 11:56:39 PM EEST
> # rpm -Uhv ntpd-4.2.6p5-5.i686.rpm
[...]
> D: opening  db index       /var/lib/rpm/Conflictname 
> thread:rdonly:auto_commit mode=0x0
> D: Conflicts: ntp < 4.2.0-3                                 NO
> D: Conflicts: ntp < 4.2.0-3                                 NO
> D: Conflicts: ntp < 4.2.0-3                                 NO
> D: Conflicts: ntp < 4.2.0-3                                 NO
> D: Conflicts: ntp < 4.2.0-3                                 NO
> D: Conflicts: ntp < 4.2.0-3                                 NO
> D: Conflicts: ntp < 4.2.0-3                                 NO
> D: Conflicts: ntp < 4.2.0-3                                 NO
> D: Conflicts: ntp < 4.2.0-3                                 NO
> D: Conflicts: ntp < 4.2.0-3                                 NO
> D: Conflicts: ntp < 4.2.0-3                                 NO
> D: Conflicts: ntp < 4.2.0-3                                 NO
> D: Conflicts: ntp < 4.2.0-3                                 NO
> D: Conflicts: ntp < 4.2.0-3                                 NO
> D: Conflicts: ntp < 4.2.0-3                                 NO
> D: Conflicts: ntp < 4.2.0-3                                 NO
> D: Conflicts: ntp < 4.2.0-3                                 NO
> ^CD: Exiting on signal(0x2) ...

And so it happens that the problem had nothing to do with database, see
http://git.pld-linux.org/?p=packages/rpm.git;a=commit;h=d2ec8f01d4cbb82f79d8abc51f26506a58162d9c
for gory details.

-- 
Jan R?korajski                                 | PLD/Linux
SysAdm                                         | http://www.pld-linux.org/
baggins<at>mimuw.edu.pl
baggins<at>pld-linux.org


From n3npq at me.com  Fri Nov 30 14:57:23 2012
From: n3npq at me.com (Jeffrey Johnson)
Date: Fri, 30 Nov 2012 08:57:23 -0500
Subject: rpm-5.4.10-22.i686 loops forever when installing
 ntpd-4.2.6p5-5.i686.rpm
In-Reply-To: <20121130124153.GG29634@sith.mimuw.edu.pl>
References: <50846358.8090607@pld-linux.org>
 <20121130124153.GG29634@sith.mimuw.edu.pl>
Message-ID: <A5007971-5A9D-4C3D-B7D4-EADF5462197F@me.com>


On Nov 30, 2012, at 7:41 AM, Jan R?korajski <baggins at pld-linux.org> wrote:

> 
> And so it happens that the problem had nothing to do with database, see
> http://git.pld-linux.org/?

Well it has nothing to do with Berkeley DB, but ?

> p=packages/rpm.git;a=commit;h=d2ec8f01d4cbb82f79d8abc51f26506a58162d9c
> for gory details.
> 

? if the attached patch to add the namespace comparison
is "fixing", then the error message is hiding the fact
that some namespace is matching the conflict unexpectedly
erroneously.

Good: there is a fix. I thought this had been patched already?
Were there other locations that needed to include the namespace
comparison?

The actual namespace and the intended semantic should be identified also:
PLD is doing something different than other distros.

Nothing wrong with different than other distros at all: I am
trying to understand what is needed before bug claims start.

73 de Jeff


> -- 
> Jan R?korajski                                 | PLD/Linux
> SysAdm                                         | http://www.pld-linux.org/
> baggins<at>mimuw.edu.pl
> baggins<at>pld-linux.org
> _______________________________________________
> pld-devel-en mailing list
> pld-devel-en at lists.pld-linux.org
> http://lists.pld-linux.org/mailman/listinfo/pld-devel-en



From n3npq at me.com  Fri Nov 30 15:11:38 2012
From: n3npq at me.com (Jeffrey Johnson)
Date: Fri, 30 Nov 2012 09:11:38 -0500
Subject: rpm-5.4.10-22.i686 loops forever when installing
 ntpd-4.2.6p5-5.i686.rpm
In-Reply-To: <A5007971-5A9D-4C3D-B7D4-EADF5462197F@me.com>
References: <50846358.8090607@pld-linux.org>
 <20121130124153.GG29634@sith.mimuw.edu.pl>
 <A5007971-5A9D-4C3D-B7D4-EADF5462197F@me.com>
Message-ID: <1ACD5758-0A5C-496E-A453-CEE7EC2C4ACE@me.com>


On Nov 30, 2012, at 8:57 AM, Jeffrey Johnson <n3npq at me.com> wrote:

> 
> The actual namespace and the intended semantic should be identified also:
> PLD is doing something different than other distros.
> 

From git log
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.


There are 2 semantics for the namespace user(?) that will surely
cause additional issues. Something has to change to end (or unify)
the confused semantics.

What actions in PLD need the virtual provide
	Provide: user(ntp)

Can a different string choice be substituted?

73 de Jeff



From baggins at pld-linux.org  Fri Nov 30 20:43:02 2012
From: baggins at pld-linux.org (Jan =?utf-8?Q?R=C4=99korajski?=)
Date: Fri, 30 Nov 2012 20:43:02 +0100
Subject: rpm-5.4.10-22.i686 loops forever when installing
 ntpd-4.2.6p5-5.i686.rpm
In-Reply-To: <1ACD5758-0A5C-496E-A453-CEE7EC2C4ACE@me.com>
References: <50846358.8090607@pld-linux.org>
 <20121130124153.GG29634@sith.mimuw.edu.pl>
 <A5007971-5A9D-4C3D-B7D4-EADF5462197F@me.com>
 <1ACD5758-0A5C-496E-A453-CEE7EC2C4ACE@me.com>
Message-ID: <20121130194302.GA1688@home.lan>

On Fri, 30 Nov 2012, Jeffrey Johnson wrote:

> 
> On Nov 30, 2012, at 8:57 AM, Jeffrey Johnson <n3npq at me.com> wrote:
> 
> > 
> > The actual namespace and the intended semantic should be identified also:
> > PLD is doing something different than other distros.
> > 
> 
> From git log
> 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.
> 
> 
> There are 2 semantics for the namespace user(?) that will surely
> cause additional issues. Something has to change to end (or unify)
> the confused semantics.
> 
> What actions in PLD need the virtual provide
> 	Provide: user(ntp)

In short (quoting Jacek Konieczny <jajcus at jajcus.net>):

The same 'group(mpd)' may be provided by multiple packages (probably not
much sense with the 'mpd' group, but important for other cases) and the
group will be removed only when the last package which provides it is
removed. So the 'Provides: user(*)' and 'Provides: group(*)' are
critical for our %{user,group}{add,remove}  macros.

More here:
http://lists.pld-linux.org/mailman/pipermail/pld-devel-en/2012-September/023057.html

> Can a different string choice be substituted?

No can do, changing that will break a lot of already running systems.

-- 
Jan R?korajski                                 | PLD/Linux
SysAdm                                         | http://www.pld-linux.org/
baggins<at>mimuw.edu.pl
baggins<at>pld-linux.org


From baggins at pld-linux.org  Fri Nov 30 20:53:08 2012
From: baggins at pld-linux.org (Jan =?utf-8?Q?R=C4=99korajski?=)
Date: Fri, 30 Nov 2012 20:53:08 +0100
Subject: rpm-5.4.10-22.i686 loops forever when installing
 ntpd-4.2.6p5-5.i686.rpm
In-Reply-To: <A5007971-5A9D-4C3D-B7D4-EADF5462197F@me.com>
References: <50846358.8090607@pld-linux.org>
 <20121130124153.GG29634@sith.mimuw.edu.pl>
 <A5007971-5A9D-4C3D-B7D4-EADF5462197F@me.com>
Message-ID: <20121130195308.GB1688@home.lan>

On Fri, 30 Nov 2012, Jeffrey Johnson wrote:

> 
> On Nov 30, 2012, at 7:41 AM, Jan R?korajski <baggins at pld-linux.org> wrote:
> 
> > 
> > And so it happens that the problem had nothing to do with database, see
> > http://git.pld-linux.org/?
> 
> Well it has nothing to do with Berkeley DB, but ?
> 
> > p=packages/rpm.git;a=commit;h=d2ec8f01d4cbb82f79d8abc51f26506a58162d9c
> > for gory details.
> > 
> 
> ? if the attached patch to add the namespace comparison
> is "fixing", then the error message is hiding the fact
> that some namespace is matching the conflict unexpectedly
> erroneously.

The thing is that all dependecies get split into <Name, nsType> tuples
(foo(bar) becomes <bar, RPMNS_TYPE_foo> and _rpmtsCheck was using
only the Name half from this tuple for all tests.

The real problem that I was afraid to touch to avoid messing things up
is lib/rpmal.c:rpmalAllSatisfiesDepend() which seems to work on global
lists instead of private copies which caused the infinite loop because
it decremented interator in rpmds 'provides' list for a transaction.

> Good: there is a fix. I thought this had been patched already?
> Were there other locations that needed to include the namespace
> comparison?

Didn't find any, but I was only looking at this particular problem.

-- 
Jan R?korajski                                 | PLD/Linux
SysAdm                                         | http://www.pld-linux.org/
baggins<at>mimuw.edu.pl
baggins<at>pld-linux.org


From n3npq at me.com  Fri Nov 30 21:04:04 2012
From: n3npq at me.com (Jeffrey Johnson)
Date: Fri, 30 Nov 2012 15:04:04 -0500
Subject: rpm-5.4.10-22.i686 loops forever when installing
 ntpd-4.2.6p5-5.i686.rpm
In-Reply-To: <20121130194302.GA1688@home.lan>
References: <50846358.8090607@pld-linux.org>
 <20121130124153.GG29634@sith.mimuw.edu.pl>
 <A5007971-5A9D-4C3D-B7D4-EADF5462197F@me.com>
 <1ACD5758-0A5C-496E-A453-CEE7EC2C4ACE@me.com> <20121130194302.GA1688@home.lan>
Message-ID: <AE75BBB2-BE11-4F55-AF04-7BBFAD50E9A1@me.com>


On Nov 30, 2012, at 2:43 PM, Jan R?korajski wrote:

> On Fri, 30 Nov 2012, Jeffrey Johnson wrote:
> 
>> 
>> On Nov 30, 2012, at 8:57 AM, Jeffrey Johnson <n3npq at me.com> wrote:
>> 
>>> 
>>> The actual namespace and the intended semantic should be identified also:
>>> PLD is doing something different than other distros.
>>> 
>> 
>> From git log
>> 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.
>> 
>> 
>> There are 2 semantics for the namespace user(?) that will surely
>> cause additional issues. Something has to change to end (or unify)
>> the confused semantics.
>> 
>> What actions in PLD need the virtual provide
>> 	Provide: user(ntp)
> 
> In short (quoting Jacek Konieczny <jajcus at jajcus.net>):
> 
> The same 'group(mpd)' may be provided by multiple packages (probably not
> much sense with the 'mpd' group, but important for other cases) and the
> group will be removed only when the last package which provides it is
> removed. So the 'Provides: user(*)' and 'Provides: group(*)' are
> critical for our %{user,group}{add,remove}  macros.
> 

OK, so _SOMETHING_ is using the Provides: as a reference count.

What is that _SOMETHING_? A script that invokes ??? to query an rpmdb???
Surely there's better/other ways to ensure that a group is removed on last erase.

Here's 2 possibilities:

And there's really little that is wrong with just relying on useradd/groupadd instead
of attempting to implicitly use a Provides: existence test.

For starters, rpm headers explicitly track all user/group assignments to files in metadata.
The Seqno index is persistent, and a counter for each user/group could be maintained.

> More here:
> http://lists.pld-linux.org/mailman/pipermail/pld-devel-en/2012-September/023057.html
> 

Yes I thought I was repeating myself. As long as there are two semantics for
the user(?) and group(?) namespaces, I'm doomed to repeat myself yet again.

>> Can a different string choice be substituted?
> 
> No can do, changing that will break a lot of already running systems.
> 

*shrug*

Then we wait for the next occurrence all over again again again.

BTW, the patch you did is overly strict. Instead of drilling
the NS type everywhere, pass the "foo(bar)" original string around instead.

This is equivalent to ripping out "run-time probe" dependencies everywhere,
a patch for a feature regression I have no further interest in.

73 de Jeff
> -- 
> Jan R?korajski                                 | PLD/Linux
> SysAdm                                         | http://www.pld-linux.org/
> baggins<at>mimuw.edu.pl
> baggins<at>pld-linux.org
> _______________________________________________
> pld-devel-en mailing list
> pld-devel-en at lists.pld-linux.org
> http://lists.pld-linux.org/mailman/listinfo/pld-devel-en



From n3npq at me.com  Fri Nov 30 21:10:57 2012
From: n3npq at me.com (Jeffrey Johnson)
Date: Fri, 30 Nov 2012 15:10:57 -0500
Subject: rpm-5.4.10-22.i686 loops forever when installing
 ntpd-4.2.6p5-5.i686.rpm
In-Reply-To: <20121130195308.GB1688@home.lan>
References: <50846358.8090607@pld-linux.org>
 <20121130124153.GG29634@sith.mimuw.edu.pl>
 <A5007971-5A9D-4C3D-B7D4-EADF5462197F@me.com> <20121130195308.GB1688@home.lan>
Message-ID: <B174647C-2983-493A-BD47-B76811FA4EC7@me.com>


On Nov 30, 2012, at 2:53 PM, Jan R?korajski wrote:

> On Fri, 30 Nov 2012, Jeffrey Johnson wrote:
> 
> The real problem that I was afraid to touch to avoid messing things up
> is lib/rpmal.c:rpmalAllSatisfiesDepend() which seems to work on global
> lists instead of private copies which caused the infinite loop because
> it decremented interator in rpmds 'provides' list for a transaction.
> 

If this is the real problem, then the fix for infinite looping
has little to do with namespaces at all.

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.

Either the low level search has to save/restore loop indexes,
or the high level loop needs its own copy.

Even better:

The rpmdsNext() has never been usefully implemented with
an iterator index in private memory. Not hard to fix, just everyone
would _PHREAK_ if anything were ever really changed in RPM
for some useful/intelligent reason.

*shrug*

73 de Jeff



From draenog at pld-linux.org  Fri Nov 30 21:26:31 2012
From: draenog at pld-linux.org (Kacper Kornet)
Date: Fri, 30 Nov 2012 21:26:31 +0100
Subject: Why vim depends on vim-rt
Message-ID: <20121130202631.GA25853@camk.edu.pl>


Description to vim-rt says:

This package contains macros, documentation, syntax configuration and
manual pages for Vim. If you want to take advantage of Vim more
powerful features, you should install this package.


But I've been just surprised that it is actually required by vim? Does
someone know the reason? For test I installed vim without vim-rt and it
seems to work. But maybe I'm missing something.

-- 
  Kacper


From baggins at pld-linux.org  Fri Nov 30 21:47:13 2012
From: baggins at pld-linux.org (Jan =?utf-8?Q?R=C4=99korajski?=)
Date: Fri, 30 Nov 2012 21:47:13 +0100
Subject: rpm-5.4.10-22.i686 loops forever when installing
 ntpd-4.2.6p5-5.i686.rpm
In-Reply-To: <B174647C-2983-493A-BD47-B76811FA4EC7@me.com>
References: <50846358.8090607@pld-linux.org>
 <20121130124153.GG29634@sith.mimuw.edu.pl>
 <A5007971-5A9D-4C3D-B7D4-EADF5462197F@me.com>
 <20121130195308.GB1688@home.lan>
 <B174647C-2983-493A-BD47-B76811FA4EC7@me.com>
Message-ID: <20121130204713.GC1688@home.lan>

On Fri, 30 Nov 2012, Jeffrey Johnson wrote:

> 
> On Nov 30, 2012, at 2:53 PM, Jan R?korajski wrote:
> 
> > On Fri, 30 Nov 2012, Jeffrey Johnson wrote:
> > 
> > The real problem that I was afraid to touch to avoid messing things up
> > is lib/rpmal.c:rpmalAllSatisfiesDepend() which seems to work on global
> > lists instead of private copies which caused the infinite loop because
> > it decremented interator in rpmds 'provides' list for a transaction.
> > 
> 
> If this is the real problem, then the fix for infinite looping
> has little to do with namespaces at all.
> 
> 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.

Thanks for clarification, I was thinking I must be seeing things ;)

> Either the low level search has to save/restore loop indexes,
> or the high level loop needs its own copy.

The real fix for the loop bug below.

BTW, did you look into the problem with triggers arguments?


--- rpm-5.4.10/lib/rpmal.c~	2012-11-30 13:20:59.087249280 +0100
+++ rpm-5.4.10/lib/rpmal.c	2012-11-30 21:20:20.856611761 +0100
@@ -403,6 +403,7 @@
     const char * KName;
     availablePackage alp;
     int rc;
+    int savedIx;
 
     if (keyp) *keyp = RPMAL_NOMATCH;
 
@@ -450,10 +451,12 @@
 	if (alp->provides != NULL)	/* XXX can't happen */
 	switch (match->type) {
 	case IET_PROVIDES:
+	    savedIx = rpmdsIx(alp->provides);
 	    /* XXX single step on rpmdsNext to regenerate DNEVR string */
 	    (void) rpmdsSetIx(alp->provides, match->entryIx - 1);
 	    if (rpmdsNext(alp->provides) >= 0)
 		rc = rpmdsCompare(alp->provides, ds);
+	    (void) rpmdsSetIx(alp->provides, savedIx);
 
 	    if (rc)
 		rpmdsNotify(ds, _("(added provide)"), 0);


-- 
Jan R?korajski                                 | PLD/Linux
SysAdm                                         | http://www.pld-linux.org/
baggins<at>mimuw.edu.pl
baggins<at>pld-linux.org


From n3npq at me.com  Fri Nov 30 23:32:15 2012
From: n3npq at me.com (Jeffrey Johnson)
Date: Fri, 30 Nov 2012 17:32:15 -0500
Subject: rpm-5.4.10-22.i686 loops forever when installing
 ntpd-4.2.6p5-5.i686.rpm
In-Reply-To: <20121130204713.GC1688@home.lan>
References: <50846358.8090607@pld-linux.org>
 <20121130124153.GG29634@sith.mimuw.edu.pl>
 <A5007971-5A9D-4C3D-B7D4-EADF5462197F@me.com> <20121130195308.GB1688@home.lan>
 <B174647C-2983-493A-BD47-B76811FA4EC7@me.com> <20121130204713.GC1688@home.lan>
Message-ID: <C6A424EA-8C0E-4824-A6B8-9AF986EBD049@me.com>


On Nov 30, 2012, at 3:47 PM, Jan R?korajski <baggins at pld-linux.org> wrote:

>> 
>> If this is the real problem, then the fix for infinite looping
>> has little to do with namespaces at all.
>> 
>> 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.
> 
> Thanks for clarification, I was thinking I must be seeing things ;)
> 
>> Either the low level search has to save/restore loop indexes,
>> or the high level loop needs its own copy.
> 
> The real fix for the loop bug below.
> 

Its a bit cleaner if you change the top-side to use a
private loop index.

FWIW: your patch is what RPM used to do, and was
removed while looking at OpenMP annotated dependency
solving with helgrind.

I.e. I'm more interested in getting a proper iterator
that is thread safe than sick hack-o-rounds these days.

> BTW, did you look into the problem with triggers arguments?
> 

Nope. I have yet to be convinced of what the root
cause is: all you have been telling me is that
	RPM5 is buggy
and I'm pretty sure the final solution has yet to be seen.

73 de Jeff

> 
> --- rpm-5.4.10/lib/rpmal.c~	2012-11-30 13:20:59.087249280 +0100
> +++ rpm-5.4.10/lib/rpmal.c	2012-11-30 21:20:20.856611761 +0100
> @@ -403,6 +403,7 @@
>     const char * KName;
>     availablePackage alp;
>     int rc;
> +    int savedIx;
> 
>     if (keyp) *keyp = RPMAL_NOMATCH;
> 
> @@ -450,10 +451,12 @@
> 	if (alp->provides != NULL)	/* XXX can't happen */
> 	switch (match->type) {
> 	case IET_PROVIDES:
> +	    savedIx = rpmdsIx(alp->provides);
> 	    /* XXX single step on rpmdsNext to regenerate DNEVR string */
> 	    (void) rpmdsSetIx(alp->provides, match->entryIx - 1);
> 	    if (rpmdsNext(alp->provides) >= 0)
> 		rc = rpmdsCompare(alp->provides, ds);
> +	    (void) rpmdsSetIx(alp->provides, savedIx);
> 
> 	    if (rc)
> 		rpmdsNotify(ds, _("(added provide)"), 0);
> 
> 
> -- 
> Jan R?korajski                                 | PLD/Linux
> SysAdm                                         | http://www.pld-linux.org/
> baggins<at>mimuw.edu.pl
> baggins<at>pld-linux.org
> _______________________________________________
> pld-devel-en mailing list
> pld-devel-en at lists.pld-linux.org
> http://lists.pld-linux.org/mailman/listinfo/pld-devel-en