FAQ Search Today's Posts Mark Forums Read
» Video Reviews

» Linux Archive

Linux-archive is a website aiming to archive linux email lists and to make them easily accessible for linux users/developers.


» Sponsor

» Partners

» Sponsor

Go Back   Linux Archive > Gentoo > Gentoo Development

 
 
LinkBack Thread Tools
 
Old 06-13-2008, 07:55 PM
"Santiago M. Mola"
 
Default Packages broken by phase ordering change

Hi all,

As discussed in bug #222721, portage has changed the execution order
of phases. It seems the change was introduced in portage-2.1.5 and it
makes that, when upgrading a package, pkg_postinst is run after the
old version has been removed. This breaks packages which use
has_version in pkg_postinst to detect upgrades/downgrades. It can also
break packages in more subtle ways.

The following ebuilds are affected by has_version problem. There may
be some affected ebuilds missing, and also ebuilds broken in a
different way.

app-pda/libopensync-0.22
app-portage/conf-update-1.0
dev-libs/libotf-0.9.4
dev-libs/libotf-0.9.5
dev-libs/libotf-0.9.6
dev-libs/libotf-0.9.7
dev-util/scons-0.97
dev-util/scons-0.98.3
dev-util/scons-0.98.4
mail-filter/dspam-3.8.0-r11
media-gfx/splashutils-1.5.2.1
media-gfx/splashutils-1.5.3.4
media-gfx/splashutils-1.5.4
media-gfx/splashutils-1.5.4.1
media-gfx/splashutils-1.5.4-r1
media-libs/libdvbpsi-0.1.5
media-libs/libdvbpsi-0.1.6
media-libs/libexif-0.6.16
media-libs/libexif-0.6.16-r1
media-libs/pdflib-7.0.2
media-libs/pdflib-7.0.2_p8
media-plugins/vdr-epgsearch-0.9.19
media-plugins/vdr-epgsearch-0.9.20
media-plugins/vdr-epgsearch-0.9.21
media-plugins/vdr-epgsearch-0.9.22
media-plugins/vdr-epgsearch-0.9.23
media-plugins/vdr-epgsearch-0.9.24
media-plugins/vdr-epgsearch-0.9.24_beta19
media-plugins/vdr-epgsearch-0.9.24_beta22
media-plugins/vdr-epgsearch-0.9.24_beta23
media-plugins/vdr-epgsearch-0.9.24_beta26
media-plugins/vdr-epgsearch-0.9.24_rc2
media-tv/gentoo-vdr-scripts-0.4.0
media-tv/gentoo-vdr-scripts-0.4.1
media-tv/gentoo-vdr-scripts-0.4.2
media-tv/gentoo-vdr-scripts-0.4.3
media-tv/gentoo-vdr-scripts-0.4.3-r1
media-tv/gentoo-vdr-scripts-0.4.4
media-tv/vdrplugin-rebuild-0.2
media-video/vdr-1.4.6
media-video/vdr-1.4.7-r10
media-video/vdr-1.6.0
media-video/vdr-1.6.0_p1
media-video/vdr-1.6.0_p1-r1
media-video/vdr-1.6.0-r1
media-video/vdr-1.6.0-r2
net-analyzer/fail2ban-0.8.0-r1
net-analyzer/fail2ban-0.8.1
net-analyzer/fail2ban-0.8.2
net-dialup/ppp-2.4.4-r14
net-dialup/ppp-2.4.4-r15
net-firewall/iptables-1.3.5-r4
net-firewall/iptables-1.3.6
net-firewall/iptables-1.3.6-r1
net-firewall/iptables-1.3.7
net-firewall/iptables-1.3.8
net-firewall/iptables-1.3.8-r1
net-firewall/iptables-1.3.8-r2
net-firewall/iptables-1.3.8-r3
net-firewall/iptables-1.4.0
net-mail/getmail-4.7.6
net-mail/getmail-4.7.7
net-mail/getmail-4.7.8
net-mail/getmail-4.8.1
net-mail/mailgraph-1.14
net-misc/asterisk-1.2.13
net-misc/asterisk-1.2.13-r1
net-misc/asterisk-1.2.14
net-misc/asterisk-1.2.14-r1
net-misc/asterisk-1.2.14-r2
net-misc/asterisk-1.2.17
net-misc/asterisk-1.2.17-r1
net-misc/asterisk-1.2.21.1
net-misc/asterisk-1.2.21.1-r1
net-misc/asterisk-1.2.27
net-misc/freenet6-4.2.2
net-misc/freenet6-5.0
net-misc/freenet6-5.1
net-misc/ser-0.9.4
net-misc/ser-0.9.6
net-misc/ser-0.9.7
net-print/cups-1.2.12-r4
net-print/cups-1.2.12-r7
net-print/cups-1.2.12-r8
net-print/cups-1.3.7-r1
net-print/cups-1.3.7-r2
sys-cluster/util-vserver-0.30.214
sys-cluster/util-vserver-0.30.215
sys-fs/udev-114
sys-fs/udev-114-r1
sys-fs/udev-114-r2
sys-fs/udev-115
sys-fs/udev-115-r1
sys-fs/udev-115-r5
sys-fs/udev-115-r6
sys-fs/udev-116
sys-fs/udev-116-r1
sys-fs/udev-117
sys-fs/udev-118
sys-fs/udev-118-r1
sys-fs/udev-118-r2
sys-fs/udev-118-r3
sys-fs/udev-119
sys-fs/udev-119-r1
sys-fs/udev-120
sys-fs/udev-121
sys-fs/udev-122
sys-fs/udev-122-r1
sys-fs/udev-124
sys-process/vixie-cron-4.1-r10
www-client/surfraw-2.1.5
www-servers/apache-2.2.8
www-servers/apache-2.2.8-r3
www-servers/apache-2.2.8-r4

If the new phase order is staying, then all those packages should be
fixed. It's possible to use has_version in pkg_setup or other phase
and cache the result in a global variable.

Regards,
--
Santiago M. Mola
Jabber ID: cooldwind@gmail.com
--
gentoo-dev@lists.gentoo.org mailing list
 
Old 06-14-2008, 02:09 PM
Ciaran McCreesh
 
Default Packages broken by phase ordering change

On Fri, 13 Jun 2008 21:55:29 +0200
"Santiago M. Mola" <coldwind@gentoo.org> wrote:
> As discussed in bug #222721, portage has changed the execution order
> of phases. It seems the change was introduced in portage-2.1.5 and it
> makes that, when upgrading a package, pkg_postinst is run after the
> old version has been removed. This breaks packages which use
> has_version in pkg_postinst to detect upgrades/downgrades. It can also
> break packages in more subtle ways.

Given that the number of affected ebuilds is so high, I'd say Portage
should have to revert the changes...

This is an EAPI scope change, if anything. Although even then the
implications are a bit messy since you're talking the interaction of
two different EAPIs.

--
Ciaran McCreesh
 
Old 06-15-2008, 10:32 AM
Matthias Schwarzott
 
Default Packages broken by phase ordering change

On Freitag, 13. Juni 2008, Santiago M. Mola wrote:
> Hi all,
>
> As discussed in bug #222721, portage has changed the execution order
> of phases. It seems the change was introduced in portage-2.1.5 and it
> makes that, when upgrading a package, pkg_postinst is run after the
> old version has been removed. This breaks packages which use
> has_version in pkg_postinst to detect upgrades/downgrades. It can also
> break packages in more subtle ways.
>

So someone that has access permissions here: Please do fix the devmanual
http://devmanual.gentoo.org/ebuild-writing/functions/pkg_postinst/index.html


It now states:

Common pkg_postinst Tasks
The most common use for pkg_postinst is to display post-install informational
messages or warnings. Note that has_version will operate on the version that
was installed, which can be useful for selective upgrade messages.

Regards
Matthias
--
gentoo-dev@lists.gentoo.org mailing list
 
Old 06-15-2008, 10:39 AM
"Santiago M. Mola"
 
Default Packages broken by phase ordering change

On Sun, Jun 15, 2008 at 12:32 PM, Matthias Schwarzott <zzam@gentoo.org> wrote:
> On Freitag, 13. Juni 2008, Santiago M. Mola wrote:
>> Hi all,
>>
>> As discussed in bug #222721, portage has changed the execution order
>> of phases. It seems the change was introduced in portage-2.1.5 and it
>> makes that, when upgrading a package, pkg_postinst is run after the
>> old version has been removed. This breaks packages which use
>> has_version in pkg_postinst to detect upgrades/downgrades. It can also
>> break packages in more subtle ways.
>>
>
> So someone that has access permissions here: Please do fix the devmanual
> http://devmanual.gentoo.org/ebuild-writing/functions/pkg_postinst/index.html
>

https://bugs.gentoo.org/show_bug.cgi?id=226419

Regards,
--
Santiago M. Mola
Jabber ID: cooldwind@gmail.com
--
gentoo-dev@lists.gentoo.org mailing list
 
Old 06-19-2008, 01:21 AM
Chris Gianelloni
 
Default Packages broken by phase ordering change

On Sat, 2008-06-14 at 15:09 +0100, Ciaran McCreesh wrote:
> On Fri, 13 Jun 2008 21:55:29 +0200
> "Santiago M. Mola" <coldwind@gentoo.org> wrote:
> > As discussed in bug #222721, portage has changed the execution order
> > of phases. It seems the change was introduced in portage-2.1.5 and it
> > makes that, when upgrading a package, pkg_postinst is run after the
> > old version has been removed. This breaks packages which use
> > has_version in pkg_postinst to detect upgrades/downgrades. It can also
> > break packages in more subtle ways.
>
> Given that the number of affected ebuilds is so high, I'd say Portage
> should have to revert the changes...

Of course, you would. What else would we expect from you?

> This is an EAPI scope change, if anything. Although even then the
> implications are a bit messy since you're talking the interaction of
> two different EAPIs.

It seems that everything these days is an EAPI scope change. That's not
very useful for Gentoo, considering it's been quite some time since PMS
was proposed and we've not seen approval for either EAPI=0 or EAPI=1 (or
PMS, for that matter). What we have gotten is a half-assed "you can use
EAPI=1 in the tree to get these enumerated features" from the Council,
but that's nothing like acceptance of a spec. Perhaps if you spent a
little more time doing something more constructive than being an asshat
on the lists, PMS would have been approved long ago. Of course, that
doesn't mesh well with your apparent need to be a complete dick to
people, so continue on with the status quo.

--
Chris Gianelloni
Release Engineering Strategic Lead
Games Developer
 
Old 06-19-2008, 02:59 AM
Jeroen Roovers
 
Default Packages broken by phase ordering change

On Wed, 18 Jun 2008 18:21:24 -0700
Chris Gianelloni <wolf31o2@gentoo.org> wrote:

> It seems that everything these days is an EAPI scope change. That's
> not very useful for Gentoo, considering it's been quite some time
> since PMS was proposed and we've not seen approval for either EAPI=0
> or EAPI=1 (or PMS, for that matter). What we have gotten is a
> half-assed "you can use EAPI=1 in the tree to get these enumerated
> features" from the Council, but that's nothing like acceptance of a
> spec. Perhaps if you spent a little more time doing something more
> constructive than being an asshat on the lists, PMS would have been
> approved long ago. Of course, that doesn't mesh well with your
> apparent need to be a complete dick to people, so continue on with
> the status quo.

+1


Kind regards,
JeR
--
gentoo-dev@lists.gentoo.org mailing list
 
Old 06-19-2008, 05:52 AM
Ryan Hill
 
Default Packages broken by phase ordering change

On Wed, 18 Jun 2008 18:21:24 -0700
Chris Gianelloni <wolf31o2@gentoo.org> wrote:

> On Sat, 2008-06-14 at 15:09 +0100, Ciaran McCreesh wrote:
> > On Fri, 13 Jun 2008 21:55:29 +0200
> > "Santiago M. Mola" <coldwind@gentoo.org> wrote:
> > > As discussed in bug #222721, portage has changed the execution
> > > order of phases. It seems the change was introduced in
> > > portage-2.1.5 and it makes that, when upgrading a package,
> > > pkg_postinst is run after the old version has been removed. This
> > > breaks packages which use has_version in pkg_postinst to detect
> > > upgrades/downgrades. It can also break packages in more subtle
> > > ways.
> >
> > Given that the number of affected ebuilds is so high, I'd say
> > Portage should have to revert the changes...
>
> Of course, you would. What else would we expect from you?
>
> > This is an EAPI scope change, if anything. Although even then the
> > implications are a bit messy since you're talking the interaction of
> > two different EAPIs.
>
> It seems that everything these days is an EAPI scope change. That's
> not very useful for Gentoo, considering it's been quite some time
> since PMS was proposed and we've not seen approval for either EAPI=0
> or EAPI=1 (or PMS, for that matter). What we have gotten is a
> half-assed "you can use EAPI=1 in the tree to get these enumerated
> features" from the Council, but that's nothing like acceptance of a
> spec. Perhaps if you spent a little more time doing something more
> constructive than being an asshat on the lists, PMS would have been
> approved long ago. Of course, that doesn't mesh well with your
> apparent need to be a complete dick to people, so continue on with
> the status quo.

I don't want to start yet another hundred post thread here[i], but our
etiquette policy applies to everyone here, and I would have hoped that
as a senior developer you could at least try to take the high road and
set an example.

I know this is hypocritical coming from someone who recently called you
a giant flaming asshole, but I've been trying hard since to be more
civil because I realize that kind of behaviour is unacceptable and
nonconstructive (and again I apologize).

I'm not picking you out here, this applies to all of us (you too
Ciaran). I mean c'mon, let's quit the bitching and get shit done
already.

[i] IOW don't reply to this mail please :P


--
gcc-porting, by design, by neglect
treecleaner, for a fact or just for effect
wxwidgets @ gentoo EFFD 380E 047A 4B51 D2BD C64F 8AA8 8346 F9A4 0662
 
Old 06-19-2008, 05:57 AM
David Leverton
 
Default Packages broken by phase ordering change

On Thursday 19 June 2008 02:21:24 Chris Gianelloni wrote:
> It seems that everything these days is an EAPI scope change.

Everything change that has the potential to break existing packages, or to
make new packages incompatible with existing package managers, is an EAPI
scope change. That is the very purpose of EAPI.

> Perhaps if you spent a little more time doing something more constructive
> than being an asshat on the lists, PMS would have been approved long ago.
> Of course, that doesn't mesh well with your apparent need to be a complete
> dick to people, so continue on with the status quo.

This thread was entirely technical until now. Are such attacks really
necessary?
--
gentoo-dev@lists.gentoo.org mailing list
 

Thread Tools




All times are GMT. The time now is 06:58 AM.

VBulletin, Copyright ©2000 - 2014, Jelsoft Enterprises Ltd.
Content Relevant URLs by vBSEO ©2007, Crawlability, Inc.
Copyright 2007 - 2008, www.linux-archive.org