[gentoo-user] Question about eix-test-obsolete output

[gentoo-user] Question about eix-test-obsolete output

[Dale]

Hi,

I ran eix-test-obsolete and cleaned up a lot of the things in the 
output.  This one part, stumps me.  Just what exactly is it trying to 
tell me?  Is this not in portage, not in the world file or what? 

Output below.  Thanks

Dale

:-)  :-) 



root@smoker / # eix-test-obsolete

< SNIP >

Installed packages with a version not in the database (or masked):

[D] kde-base/kdeaddons-docs-konq-plugins (3.5.9(3.5)@04/08/2008 -> 
3.5.8(3.5)): Documentation for the konqueror plugins from kdeaddons
[D] kde-base/kdeaddons-kfile-plugins (3.5.9(3.5)@04/08/2008 -> 
3.5.8(3.5)): kdeaddons kfile plugins
[D] kde-base/kdeaddons-meta (3.5.9(3.5)@04/08/2008 -> 3.5.8(3.5)): 
kdeaddons - merge this to pull in all kdeaddons-derived packages
[D] sys-kernel/gentoo-sources (2.6.23-r3(2.6.23-r3)@02/03/2008 
2.6.23-r8(2.6.23-r8)@02/11/2008 2.6.24-r4(2.6.24-r4)@04/08/2008 -> 
2.6.16-r13(2.6.16-r13) 2.6.19-r5(2.6.19-r5) 2.6.22-r10(2.6.22-r10) 
2.6.23-r9(2.6.23-r9) 2.6.24-r3(2.6.24-r3) 2.6.24-r4(2.6.24-r4)): Full 
sources including the Gentoo patchset for the 2.6 kernel tree
[D] x11-apps/xinit (1.0.5-r2@01/22/2008 -> [m]1.0.4 [m]1.0.5-r1 
~1.0.5-r2 ~1.0.8 ~1.0.8-r2 {debug hal minimal pam}): X Window System 
initializer
Found 5 matches.

root@smoker / #      
-- 
gentoo-user@lists.gentoo.org mailing list

Re: [gentoo-user] Question about eix-test-obsolete output

[Vaeth]

> Is this not in portage, not in the world file or what? 
> 
> Installed packages with a version not in the database (or masked):

The "database" is what is produced by update-eix, i.e. usually
the portage tree and your overlays (and perhaps "virtual" overlays).
So, as a rule, it means that you have at least one version of these
packages installed which was erased from the portage tree (and isn't
in your overlays either).
-- 
gentoo-user@lists.gentoo.org mailing list

Re: [gentoo-user] Question about eix-test-obsolete output

[Daniel Pielmeier]

  > Installed packages with a version not in the database (or masked):

As it says this apps are either not in the database or masked.

kde-base/kdeaddons-docs-konq-plugins-3.5.9 masked
kde-base/kdeaddons-kfile-plugins-3.5.9 masked
kde-base/kdeaddons-meta-3.5.9 masked
sys-kernel/gentoo-sources-2.6.23-r3 not in the tree and thus not in the 
eix-database anymore.
x11-apps/xinit-1.0.5-r2 masked

I guess portage will downgrade the masked ones if you run emerge -uND 
world, so if you want them put them in package.keywords. You can remove 
gentoo-sources from /etc/portage/package.whatever as it is not in the 
tree anymore.

Regards,

Daniel
-- 
gentoo-user@lists.gentoo.org mailing list

Re: [gentoo-user] Question about eix-test-obsolete output

[Dale]

Vaeth wrote:
>> Is this not in portage, not in the world file or what? 
>>
>> Installed packages with a version not in the database (or masked):
>>     
>
> The "database" is what is produced by update-eix, i.e. usually
> the portage tree and your overlays (and perhaps "virtual" overlays).
> So, as a rule, it means that you have at least one version of these
> packages installed which was erased from the portage tree (and isn't
> in your overlays either).
>   

Just to give a little more info here.  I sync'ed a couple days ago.  I 
use eix-sync to do that so that everything is updated, portage and eix.  
I don't have anything in layman, heard of it but never used it. 

Also, emerge -uvDNp comes out clean.  Nothing to upgrade or downgrade.  
Revdep-rebuild comes out clean as well. 

Still not having that light bulb moment here.  :/  Are these packages 
that I can unmerge?

Dale

:-)  :-) 
-- 
gentoo-user@lists.gentoo.org mailing list

Re: [gentoo-user] Question about eix-test-obsolete output

[Dale]

Daniel Pielmeier wrote:
>  > Installed packages with a version not in the database (or masked):
>
> As it says this apps are either not in the database or masked.
>
> kde-base/kdeaddons-docs-konq-plugins-3.5.9 masked
> kde-base/kdeaddons-kfile-plugins-3.5.9 masked
> kde-base/kdeaddons-meta-3.5.9 masked
> sys-kernel/gentoo-sources-2.6.23-r3 not in the tree and thus not in 
> the eix-database anymore.
> x11-apps/xinit-1.0.5-r2 masked
>
> I guess portage will downgrade the masked ones if you run emerge -uND 
> world, so if you want them put them in package.keywords. You can 
> remove gentoo-sources from /etc/portage/package.whatever as it is not 
> in the tree anymore.
>
> Regards,
>
> Daniel


emerge -uNDvp world comes out clean.  I can not find 
gentoo-sources-2.6.23-r3 in any file in /etc/portage/package.* so sort 
of curious about that.  I did check and I am not using that version 
anymore so I unmerged it.  Maybe that will fix that.  o_O

Still waiting on that light bulb moment.  I been studying this thing the 
past couple days on my own so it's a bit . . . muddy, in here.  LOL

Dale

:-)  :-) 


-- 
gentoo-user@lists.gentoo.org mailing list

Re: [gentoo-user] Question about eix-test-obsolete output

[Daniel Pielmeier]

  > emerge -uNDvp world comes out clean.

Hmm. I just commented an entry in package keywords and after that it 
showed up the same way as reported it. If i run portage it wants to 
downgrade that particular package.

> I can not find 
> gentoo-sources-2.6.23-r3 in any file in /etc/portage/package.* so sort 
> of curious about that.  I did check and I am not using that version 
> anymore so I unmerged it.  Maybe that will fix that.  o_O

This should fix it! It was my mistake. Eix is looking up its database 
and does not find gentoo-sources-2.6.23-r3 because it has been removed 
from the tree thus it gets reported.

> Still waiting on that light bulb moment.  I been studying this thing the 
> past couple days on my own so it's a bit . . . muddy, in here.  LOL
> 
> Dale
> 
> :-)  :-)
> 

@Vaeth

Maybe you can enlighten me.

eix-test-obsolete checks first for non matching entries in a package 
file e.g. if I have an entry with a typo or a removed package then it 
will be reported.

But what is the difference between a "redundant entry" and an 
"uninstalled entry". As far as I see the matching criteria of both 
checks is a package which is not installed or in the database but in a 
package.* file.

Btw: eix-test-obsolete can not check for obsolete use-flags at the moment?

Regards,

Daniel
-- 
gentoo-user@lists.gentoo.org mailing list

Re: [gentoo-user] Question about eix-test-obsolete output

[Dale]

Daniel Pielmeier wrote:
>
>
> Btw: eix-test-obsolete can not check for obsolete use-flags at the 
> moment?
>
> Regards,
>
> Daniel

If it did check USE flags, I'd have a loooong list there too.  This is a 
5 year old install.  I try to keep it tidy but it does creep up on me. 

Dale

:-)  :-) 


-- 
gentoo-user@lists.gentoo.org mailing list

Re: [gentoo-user] Question about eix-test-obsolete output

[Neil Bothwick]

[-- Attachment #1: Type: text/plain, Size: 738 bytes --]

On Sun, 13 Apr 2008 14:02:38 +0200, Daniel Pielmeier wrote:

> But what is the difference between a "redundant entry" and an 
> "uninstalled entry". As far as I see the matching criteria of both 
> checks is a package which is not installed or in the database but in a 
> package.* file.

Redundant is where the package is still available but the /etc/portage.*
entry is no longer needed. e.g. you have "dev-lib/foobar-1.1 ~x86" in
package.keyworkd but it is now stable.


Btw: eix-test-obsolete can not check for obsolete use-flags at the
> moment?

flagedit will warn if you have obsolete flags in /etc/portage
or /etc/make.conf.


-- 
Neil Bothwick

I have plenty of talent and vision. I just don't give a damn.

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 197 bytes --]

Re: [gentoo-user] Question about eix-test-obsolete output

[Daniel Pielmeier]

Neil Bothwick schrieb:
> On Sun, 13 Apr 2008 14:02:38 +0200, Daniel Pielmeier wrote:
> 
>> But what is the difference between a "redundant entry" and an 
>> "uninstalled entry". As far as I see the matching criteria of both 
>> checks is a package which is not installed or in the database but in a 
>> package.* file.
> 
> Redundant is where the package is still available but the /etc/portage.*
> entry is no longer needed. e.g. you have "dev-lib/foobar-1.1 ~x86" in
> package.keyworkd but it is now stable.

Sounds reasonable, thanks!

> Btw: eix-test-obsolete can not check for obsolete use-flags at the
>> moment?
> 
> flagedit will warn if you have obsolete flags in /etc/portage
> or /etc/make.conf.
> 

Thanks!

I wonder when there will be one single tool which is capable to take 
care of a configuration and cleaning /etc/portage/ or is there already 
one i miss?

Regards,

Daniel

-- 
gentoo-user@lists.gentoo.org mailing list

Re: [gentoo-user] Question about eix-test-obsolete output

[Dale]

Daniel Pielmeier wrote:
>
> Thanks!
>
> I wonder when there will be one single tool which is capable to take 
> care of a configuration and cleaning /etc/portage/ or is there already 
> one i miss?
>
> Regards,
>
> Daniel
>

I wouldn't mind having one that cleans out /etc as a whole.  I'm sure 
there are some old configs in there that are no longer needed.  You 
know, package is long gone but configs are still laying around.

Dale

:-)  :-)
-- 
gentoo-user@lists.gentoo.org mailing list

Re: [gentoo-user] Question about eix-test-obsolete output

[Vaeth]

Neil Bothwick wrote:
> > On Sun, 13 Apr 2008 14:02:38 +0200, Daniel Pielmeier wrote:
> >
> > Redundant is where the package is still available but the /etc/portage.*
> > entry is no longer needed. e.g. you have "dev-lib/foobar-1.1 ~x86" in
> > package.keyworkd but it is now stable.
>
> Sounds reasonable, thanks!

There are much more types of redundancies. E.g. if you marked it twice
or with two different keywords etc. Use the REDUNDANT_IF_* variables
(described on the eix manpage) to tell eix-test-obsolete in detail what
types _you_ do consider as redundant.

> > Btw: eix-test-obsolete can not check for obsolete use-flags at the
> > moment?

eix does not have support for useflags at all (except for some very
primitive cases and checks).

> I wonder when there will be one single tool [...]
> cleaning /etc/portage/

I think there will never be a convenient tool, because there are too
many cases: What somebody considers as obsolete (e.g. an ** unmasking
when meanwhile ~x86 is available) is a natural entry for another person.
IMHO, except for useflags, eix-test-obsolete cannot much be improved
for such a task. In new eix versions you can even provide a file for
packages for which you do *not* want to see warnings (e.g. if you *know*
that you will always want the ~x86 version of that package, you will
not get any warnings if the highest version is stable).

Regards
Martin
-- 
gentoo-user@lists.gentoo.org mailing list

Re: [gentoo-user] Question about eix-test-obsolete output

[Daniel Pielmeier]

Dale schrieb:
> Daniel Pielmeier wrote:
>>
>> Thanks!
>>
>> I wonder when there will be one single tool which is capable to take 
>> care of a configuration and cleaning /etc/portage/ or is there already 
>> one i miss?
>>
>> Regards,
>>
>> Daniel
>>
> 
> I wouldn't mind having one that cleans out /etc as a whole.  I'm sure 
> there are some old configs in there that are no longer needed.  You 
> know, package is long gone but configs are still laying around.
> 
> Dale
> 
> :-)  :-)

I am not a programmer, but I managed to get a bit familiar with perl. So 
I wrote myself a small cruft-script in which takes care of obsolete 
files. It searches for all files on the disk instead those which are on 
a ignorelist. This list is compared with all files managed by portage. 
Everything which is left is not needed anymore. I do not share this 
script as you have to be extremely careful which files you delete. If 
you remove files which are false positives, it can break the whole system!

What I want is one tool which does not only manage /etc/portage. It 
should also take care of the world file, as there are often many 
unneeded entries which are already dependencies of another package.
Also it should be capable of cleaning out unneeded binary packages and 
distfiles. It removes unneeded slots or packages. It takes care of 
linkage when there are abi changes.

This is all doable until now but I have about six different tools which 
do their job more or less reliable to achieve all this.

I know this are high pretensions, but such a tool would be very powerful!

Regards,

Daniel
-- 
gentoo-user@lists.gentoo.org mailing list

Re: [gentoo-user] Question about eix-test-obsolete output

[Vaeth]

Dale wrote:
> Vaeth wrote:
> > > Is this not in portage, not in the world file or what? 
> > >
> > > Installed packages with a version not in the database (or masked):
> 
> Also, emerge -uvDNp comes out clean.  Nothing to upgrade or downgrade.
> Revdep-rebuild comes out clean as well.

The installed packages are not necessarily in a one-to-one correspondence
with emerge -NaDuv world.
More precisely, the following are the exceptions:

1. It might be that a package A was installed earlier as a dependency of
   another package B. If you then uninstalled that package B, or if you
   changed a useflags of package B, or if the ebuild author changed
   the dependencies for package B, it might be that package A is no
   longer needed. So emerge -NaDuv world will not see a reason to upgrade
   this package.

2. It might be a slotted package, and an old slot is not needed by
   anything.

Ideally, you should see both types listed for removal with
emerge -p --depclean

eix does not care *why* a package is installed. If you just list the 
package in question with eix, you will see which versions are installed 
and which versions are available in the database (i.e. in your case: 
available in the portage tree), and you will see that the installed 
version (or at least one of them) is not available anymore...

Usually this is a hint that you will want to upgrade that version or
uninstall it or that you want to keep a local copy of the original ebuild
in your overlay (and usually also of the tarball) (the latter only
for the case that you disagree with the gentoo developers and want to
continue to use the old version for some reason - you have to make sure
that you can still reemerge the package for the case that e.g.
revdep-rebuilds finds a missing library in that package).

Regards
Martin
-- 
gentoo-user@lists.gentoo.org mailing list

Re: [gentoo-user] Question about eix-test-obsolete output

[Daniel Pielmeier]

Vaeth schrieb:
> Neil Bothwick wrote:
>>> On Sun, 13 Apr 2008 14:02:38 +0200, Daniel Pielmeier wrote:
>>>
>>> Redundant is where the package is still available but the /etc/portage.*
>>> entry is no longer needed. e.g. you have "dev-lib/foobar-1.1 ~x86" in
>>> package.keyworkd but it is now stable.
>> Sounds reasonable, thanks!
> 
> There are much more types of redundancies. E.g. if you marked it twice
> or with two different keywords etc. Use the REDUNDANT_IF_* variables
> (described on the eix manpage) to tell eix-test-obsolete in detail what
> types _you_ do consider as redundant.

Already read about this, now I think I fully understand it!

>>> Btw: eix-test-obsolete can not check for obsolete use-flags at the
>>> moment?
> 
> eix does not have support for useflags at all (except for some very
> primitive cases and checks).
> 
>> I wonder when there will be one single tool [...]
>> cleaning /etc/portage/
> 
> I think there will never be a convenient tool, because there are too
> many cases: What somebody considers as obsolete (e.g. an ** unmasking
> when meanwhile ~x86 is available) is a natural entry for another person.
> IMHO, except for useflags, eix-test-obsolete cannot much be improved
> for such a task. In new eix versions you can even provide a file for
> packages for which you do *not* want to see warnings (e.g. if you *know*
> that you will always want the ~x86 version of that package, you will
> not get any warnings if the highest version is stable).

Thank you Martin this is very useful information! I always wondered if 
the output could be suppressed for special packages.

I also see now that taking care of /etc/portage/ is a very difficult job!

You are thinking about implementing better support for use-flags?

Regards,

Daniel
-- 
gentoo-user@lists.gentoo.org mailing list

Re: [gentoo-user] Question about eix-test-obsolete output

[Peter Ruskin]

On Sunday 13 April 2008, Daniel Pielmeier wrote:
> What I want is one tool which does not only manage /etc/portage.
> It should also take care of the world file, as there are often
> many unneeded entries which are already dependencies of another
> package. Also it should be capable of cleaning out unneeded
> binary packages and distfiles. It removes unneeded slots or
> packages. It takes care of linkage when there are abi changes.

man emaint (comes with portage)

-- 
Peter
========================================================================
Gentoo Linux: Portage 2.1.5_rc2		kernel-2.6.23-gentoo-r3
AMD Athlon(tm) 64 X2 Dual Core Processor 4400+	gcc(GCC): 4.1.2
KDE: 3.5.8					Qt: 3.3.8
========================================================================
-- 
gentoo-user@lists.gentoo.org mailing list

Re: [gentoo-user] Question about eix-test-obsolete output

[Neil Bothwick]

[-- Attachment #1: Type: text/plain, Size: 649 bytes --]

On Sun, 13 Apr 2008 17:36:07 +0200, Daniel Pielmeier wrote:

> > flagedit will warn if you have obsolete flags in /etc/portage
> > or /etc/make.conf.

> I wonder when there will be one single tool which is capable to take 
> care of a configuration and cleaning /etc/portage/ or is there already 
> one i miss?

That would be one huge, bloated program. Using separate tools for
separate jobs is the *nix way of doing things, and flagedit and eix have
very different roles. eix provides information, whereas flagedit is a
configuration editor.


-- 
Neil Bothwick

"Bother," said Pooh, more from force of habit than anything else.

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 197 bytes --]

Re: [gentoo-user] Question about eix-test-obsolete output

[Neil Bothwick]

[-- Attachment #1: Type: text/plain, Size: 474 bytes --]

On Sun, 13 Apr 2008 19:25:52 +0200, Daniel Pielmeier wrote:

> This is all doable until now but I have about six different tools which 
> do their job more or less reliable to achieve all this.

Having separate tools has the advantage that it is possible to improve
or replace individual ones, such as eix superceding esearch, without
having to worry about how all the other functions are handled.


-- 
Neil Bothwick

Top Oxymorons Number 21: "Now, then ..."

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 197 bytes --]