[gentoo-dev] Cruft detecting script

[gentoo-dev] Cruft detecting script

[Andy Arbon]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Hello,

I've been wondering about writing a cruft detecting script.. The basic
idea would be:
* Produce a list of every file that Portage claims responsibility for on
the current system (ie. the results of qpkg -l `qpkg -I`)
* Produce a list of every file in the system (ie. find /)
* Remove the first set from the second
* Whatever's left is 'cruft'

The above needs to be tweaked so that find doesn't look in areas that
are not used for system things, eg. /home, /tmp etc.

Has anyone written something like this, and does anyone have any reason
why something like this wouldn't work or be useful?

Cheers,

Andy

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.2.1 (GNU/Linux)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org

iD8DBQE+YjTeX3TTUvZURBERAhcmAKCgpIMVRugO9gbeKvTKG+TgObI0pgCeO4fb
bRqkUTm51aLUB8/Lpn5J1BQ=
=4xNI
-----END PGP SIGNATURE-----


--
gentoo-dev@gentoo.org mailing list

Re: [gentoo-dev] Cruft detecting script

[Brian Jackson]

On Sunday 02 March 2003 10:44 am, Andy Arbon wrote:
> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
>
> Hello,
>
> I've been wondering about writing a cruft detecting script.. The basic
> idea would be:
> * Produce a list of every file that Portage claims responsibility for on
> the current system (ie. the results of qpkg -l `qpkg -I`)
> * Produce a list of every file in the system (ie. find /)
> * Remove the first set from the second
> * Whatever's left is 'cruft'
>
> The above needs to be tweaked so that find doesn't look in areas that
> are not used for system things, eg. /home, /tmp etc.
>
> Has anyone written something like this, and does anyone have any reason
> why something like this wouldn't work or be useful?

I think this could be very useful to some, but not very useful for others. I 
often install software not using an ebuild, so for me it would detect some of 
the stuff that I installed without an ebuild. 

--Brian

>
> Cheers,
>
> Andy
>
> -----BEGIN PGP SIGNATURE-----
> Version: GnuPG v1.2.1 (GNU/Linux)
> Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org
>
> iD8DBQE+YjTeX3TTUvZURBERAhcmAKCgpIMVRugO9gbeKvTKG+TgObI0pgCeO4fb
> bRqkUTm51aLUB8/Lpn5J1BQ=
> =4xNI
> -----END PGP SIGNATURE-----


--
gentoo-dev@gentoo.org mailing list

Re: [gentoo-dev] Cruft detecting script

[Mark Gordon]

On Sun, 2 Mar 2003 11:12:40 -0600
Brian Jackson <brian@mdrx.com> wrote:

> On Sunday 02 March 2003 10:44 am, Andy Arbon wrote:

> > I've been wondering about writing a cruft detecting script.. The
> > basic idea would be:
> > * Produce a list of every file that Portage claims responsibility
> > for on the current system (ie. the results of qpkg -l `qpkg -I`)
> > * Produce a list of every file in the system (ie. find /)
> > * Remove the first set from the second
> > * Whatever's left is 'cruft'
> >
> > The above needs to be tweaked so that find doesn't look in areas
> > that are not used for system things, eg. /home, /tmp etc.
> >
> > Has anyone written something like this, and does anyone have any
> > reason why something like this wouldn't work or be useful?
> 
> I think this could be very useful to some, but not very useful for
> others. I often install software not using an ebuild, so for me it
> would detect some of the stuff that I installed without an ebuild. 

Exclude /usr/local as well. That should exclude most software not
installed through portage. Possibly use a config file so that other
directories can be excluded by the user.

I think it would be useful for people who know what they are doing,
dangerous for people who only *think* they know what they are doing.
-- 
Mark Gordon
Paid to be a Geek & a Senior Software Developer
Currently looking for a new job commutable from Slough, Berks, U.K.
Although my email address says spamtrap, it is real and I read it.

--
gentoo-dev@gentoo.org mailing list

[gentoo-dev] Cruft detecting script

[Ian Phillips]

> Exclude /usr/local as well. That should exclude most software not
> installed through portage. Possibly use a config file so that other
> directories can be excluded by the user.

It definitely (for me anyway, and probably as a general principle) needs
a user configurable exclude list. For the default excludes I'd suggest

/home
/tmp
/usr/local
/usr/portage
/var/tmp

I've included /usr/portage since portage won't know about these files
even though it manages them. There's a script floating around somewhere that
cleans all of the old files out of your /usr/portage/distfiles directory,
and this sounds like it would make a nice complement to that. Maybe they
could both be included in a later version of gentoolkit.

Yours,
Ian.

-- 
#ifndef  __COMMON_SENSE__ | Ian Phillips
#include <std_disclaimer> | TIBCO Software Inc.
#endif                    | www.tibco.com




--
gentoo-dev@gentoo.org mailing list

Re: [gentoo-dev] Cruft detecting script

[Gareth John]

>
> Exclude /usr/local as well. That should exclude most software not
> installed through portage. Possibly use a config file so that other
> directories can be excluded by the user.
>

You should probably also exclude any NFS-mounted partitions aswell...

Gareth




--
gentoo-dev@gentoo.org mailing list

RE: [gentoo-dev] Cruft detecting script

[Sami Dalouche]

Hi,

It seems everybody agrees to make a configurable exclude-list for your 
Script.
What I suggest is that you take the debian "cruft" package, and modify 
It so that it works with gentoo.
The way it works is easy :
 1) produces a list of every file in the system (find /)
 2) runs every script in /etc/cruft/exclude/ (or whatever)
 3) remove 2 from 1
 4) remove all package-owned files from 1)
 5) print the rest

As a result, you can configure the files you want excluded by creating
simple
#!/bin/sh
find /usr/local

Scripts, for example..


Sam

> -----Original Message-----
> From: Andy Arbon [mailto:gentoo@andrewarbon.co.uk] 
> Sent: dimanche 2 mars 2003 17:44
> To: gentoo-dev
> Subject: [gentoo-dev] Cruft detecting script
> 
> 
> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
> 
> Hello,
> 
> I've been wondering about writing a cruft detecting script.. 
> The basic idea would be:
> * Produce a list of every file that Portage claims 
> responsibility for on the current system (ie. the results of 
> qpkg -l `qpkg -I`)
> * Produce a list of every file in the system (ie. find /)
> * Remove the first set from the second
> * Whatever's left is 'cruft'
> 
> The above needs to be tweaked so that find doesn't look in 
> areas that are not used for system things, eg. /home, /tmp etc.
> 
> Has anyone written something like this, and does anyone have 
> any reason why something like this wouldn't work or be useful?
> 
> Cheers,
> 
> Andy
> 
> -----BEGIN PGP SIGNATURE-----
> Version: GnuPG v1.2.1 (GNU/Linux)
> Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org
> 
> iD8DBQE+YjTeX3TTUvZURBERAhcmAKCgpIMVRugO9gbeKvTKG+TgObI0pgCeO4fb
> bRqkUTm51aLUB8/Lpn5J1BQ=
> =4xNI
> -----END PGP SIGNATURE-----
> 
> 
> --
> gentoo-dev@gentoo.org mailing list
> 
> 
> 


--
gentoo-dev@gentoo.org mailing list

Re: [gentoo-dev] Cruft detecting script

[Evan Powers]

On Sunday 02 March 2003 11:44 am, Andy Arbon wrote:
> Has anyone written something like this, and does anyone have any reason
> why something like this wouldn't work or be useful?

I was bored so I quickly hacked this together, just to see how it would do.

time sh script-cruft.sh
...lots of output....
real    0m48.147s
user    0m33.010s
sys     0m3.810s

My first attempt was more like "real 125m48.147s", but I just wasn't very 
intelligent about that. This version actually exploits sed's capabilities. A 
15577% increase in performance for a few minutes of work isn't too shabby, if 
I may say so myself. ;-)

The script produced far more output than I would have naively expected, about 
4300 lines. There might be too much noise for the tool to be useful. On the 
other hand....

It did catch some things that surprised me. On my system there are lots of 
libraries and perl related files that aren't known to portage (though they 
correspond to software I've installed via an ebuild). For example, on my 
system there are several programs in /usr/bin which portage doesn't know 
about.

/usr/bin/amstex -> tex
/usr/bin/cont-cz -> pdfetex
/usr/bin/cont-de -> pdfetex
/usr/bin/cont-en -> pdfetex
/usr/bin/cont-nl -> pdfetex
/usr/bin/cont-uk -> pdfetex
/usr/bin/cpp
/usr/bin/cpp.orig
/usr/bin/elatex -> etex
/usr/bin/eplain -> tex
/usr/bin/frlatex -> tex
/usr/bin/lambda -> omega
/usr/bin/latex -> tex
/usr/bin/pdfelatex -> pdfetex
/usr/bin/pdflatex -> pdftex
/usr/bin/rec -> /usr/bin/play

Makes me think the app-text/tetex-1.0.7-r11 ebuild was broken; don't know if 
the current stable 1.0.7-r12 does the same thing, I'll have to look into 
that. The /usr/bin/cpp looks like a script related to sys-devel/gcc-config, 
you'd think portage would know about this too.

---script-cruft.sh---
#!/bin/sh

find / '(' -path /proc \
	-or -path /dev \
	-or -path /boot \
	-or -path /mnt \
	-or -path /tmp \
	-or -path /var/tmp \
	-or -path /root \
	-or -path /home \
	-or -path /lib/dev-state \
	-or -path /lib/modules \
	-or -path /usr/portage \
	-or -path /var/cache/edb \
	-or -path /var/db/pkg \
	')' -prune -or -print \
| sort >/tmp/allfiles

qpkg -nc -l \
| sed -n -e 's/ -> .*//' -e '1,2 d' -e '/^$/,+2! p' \
| sort \
| uniq >/tmp/portagefiles

comm -2 -3 /tmp/allfiles /tmp/portagefiles
---script-cruft.sh---



--
gentoo-dev@gentoo.org mailing list