Re: [gentoo-user] Re: smoothest way to jump from 2006 to 2008
On Thursday 01 May 2008, Neil Bothwick wrote: On Wed, 30 Apr 2008 22:17:34 +0200, Alan McKinnon wrote: [2] vi /var/lib/portage/world Edit at will with sense of abandon vi /etc/make.conf Edit where appropriate vi /etc/portage/* Fearlessly edit throwing caution to the winds loop_entry: emerge -avuND world Resolve blockers emerge --resume --skipfirst (bonus points if you correctly predict which packages will fail to build) emerge -av --depclean goto loop_entry What a horrible suggestion, not only does it use vi - three time! - but it ends with a goto! Here's a much more horrible thing: In which way does the suggestion, horrible as it is, depart from reality? -- Alan McKinnon alan dot mckinnon at gmail dot com -- gentoo-user@lists.gentoo.org mailing list
Re: [gentoo-user] Re: smoothest way to jump from 2006 to 2008
On Thu, 1 May 2008 10:30:18 +0200, Alan McKinnon wrote: What a horrible suggestion, not only does it use vi - three times! - but it ends with a goto! Here's a much more horrible thing: In which way does the suggestion, horrible as it is, depart from reality? Your reality must be bad if you have to use vi that often ;-) -- Neil Bothwick He's so cool, he could get frostbite from masturbating. signature.asc Description: PGP signature
Re: [gentoo-user] Re: smoothest way to jump from 2006 to 2008
On Thursday 01 May 2008, [EMAIL PROTECTED] wrote: Alan McKinnon [EMAIL PROTECTED] writes: [2] vi /var/lib/portage/world What if I emerge -vC all I know I don't want. All kde all gnome all xorg for example, before update world. It seems fine in theory, but every time I've done it in real life I get into deep dependency trees that take more time to sort out than simply emerging world first. The trouble seems to be that it's easy to spot the high level packages that bring in all their DEPENDS - things like kde,gnome,xorg-x11. It's harder to spot the dependencies that want to bring everything back in again. kde is easy - the string 'kde' shows up in most of the names. gnome is harder as the names are much more randomly chosen. Plus, emerge --depclean will refuse to run until emerge world returns nothing to do. So you first have to get all the gui stuff out of the world file, identify *all* installed gui packages not in world and emerge -C them, then run emerge world followed by emerge --depclean. Like I said, I have found every time I've done it that it actually took longer and wasted more of my time and caused more frustration than just building kde. At least when building kde I could walk away and come back tomorrow. Fiddling world requires you to be there, watch the screen, decide, type next command, repeat. -- Alan McKinnon alan dot mckinnon at gmail dot com -- gentoo-user@lists.gentoo.org mailing list
Re: [gentoo-user] Re: smoothest way to jump from 2006 to 2008
On Thursday 01 May 2008, Neil Bothwick wrote: On Thu, 1 May 2008 10:30:18 +0200, Alan McKinnon wrote: What a horrible suggestion, not only does it use vi - three times! - but it ends with a goto! Here's a much more horrible thing: In which way does the suggestion, horrible as it is, depart from reality? Your reality must be bad if you have to use vi that often ;-) The last time my machine was (a) switch on for more than 5 minutes and (b) was not running vi either locally or in a remote screen session, was ... lemme think ... sometime around 2004?? vi is good vi is TheOneTrueUnixWay(tm) vi runs anywhere vi works on keyboards from the '70s vi was written by Bill Joy Bill Joy wrote lots of good stuff in BSD Our favourite OS owes a lot to BSD using vi pays homage to those magnificent BSD'ers of old If you hold up your hand with all fingers except the ring finger extended, it spells vi to others Bram works at Google I had a vi question two weeks ago. A colleague found a good friend at Google on IRC and asked him for help with the vi question. This good friend said Hang on a sec, I'll ask Bram, he sits two desks away (!!) The universe does not drop obvious nano hints in our laps like it does with vi (see above). The universe's will in this regard is self-evident. I rest my case. -- Alan McKinnon alan dot mckinnon at gmail dot com -- gentoo-user@lists.gentoo.org mailing list
[gentoo-user] Re: smoothest way to jump from 2006 to 2008
Alan McKinnon [EMAIL PROTECTED] writes: vi was written by Bill Joy Bill Joy wrote lots of good stuff in BSD Our favourite OS owes a lot to BSD using vi pays homage to those magnificent BSD'ers of old I second all the vi accolades. I like the fact that Bill Joy was horribly drunk when he wrote the bulk of vi. vi is a very powerfull editor. I'll probably cause Alan to never help me again but I will admit I mostly use vim when the OS is pretty finished. -- gentoo-user@lists.gentoo.org mailing list
[gentoo-user] Re: smoothest way to jump from 2006 to 2008
Alan McKinnon [EMAIL PROTECTED] writes:
All kde all gnome all xorg for example, before update world.
It seems fine in theory, but every time I've done it in real life I get
into deep dependency trees that take more time to sort out than simply
emerging world first.
The trouble seems to be that it's easy to spot the high level packages
that bring in all their DEPENDS - things like kde,gnome,xorg-x11. It's
harder to spot the dependencies that want to bring everything back in
again. kde is easy - the string 'kde' shows up in most of the names.
gnome is harder as the names are much more randomly chosen.
It is turning into a hefty time sink but I doubt it would take less
time to grind out update world. And in this case I want to get rid of
all x related stuff for good.
I've found (just now) that using revdep-rebuild --pretend and then
going thru the list and emerge -vC parent packages or in some cases
emerge -v them will eventually get you a pretty clean start for build
world.
But in my case it was made a very lot easier by knowing in advance I
wanted no media related gnome, kde, xorg type stuff.
It meant I could run stuff like:
for ii in $(eix -Ic |grep x11-|awk '/^\[/{print $2}' |sort)
do
emerge -avC $ii; done
done
Insert media-app, plugin, kde, gnome in the grep part and walk thru
using `ask' in case something needs to be left alone.
Then following up with several runs at revdep-rebuild as cited above.
That has taken a while.. several hours... but wouldn't such big update
world have taken at least a day?
--
gentoo-user@lists.gentoo.org mailing list
Re: [gentoo-user] Re: smoothest way to jump from 2006 to 2008
My cleanup routine pretty much involves running vim on /var/lib/portage/world and going down the list. If I see something I definitely don't need, I remove that line. If I see something that I don't remember what it was, in another terminal (just an ALT-TAB away) I run esearch package-name. After I'm done going through the file, I run emerge -a --depclean followed by revdep-rebuild. Call me old school, but that method never takes me more than a few minutes to do. I am also someone in the vim camp. It fires up quickly and is very efficient for moving around files, which is what editing configuration files is all about. Not to mention that I use it for all of my programming as well (except that I'm just now starting to learn SLIME for Lisp, pretty much a necessity). The other thing I do is to run emerge --sync and emerge -uDNav once a week (followed by emerge --depclean and revdep-rebuild). I check the USE flags of the updated/newly-installed packages to make sure they are what I want, using euse -E, euse -D, or editing /etc/portage/package.use appropriately. That's my input, Brandon -- gentoo-user@lists.gentoo.org mailing list
[gentoo-user] Re: smoothest way to jump from 2006 to 2008
Brandon Mintern [EMAIL PROTECTED] writes: Call me old school, but that method never takes me more than a few minutes to do. I am also someone in the vim camp. It fires up quickly Unless you are updating a vmappliance built on old (even for 2006) 2006 pkgs to current 2008 pkgs. I know this is a monster thread but making that big a jump is what is being discussed. -- gentoo-user@lists.gentoo.org mailing list
Re: [gentoo-user] Re: smoothest way to jump from 2006 to 2008
On Thu, May 1, 2008 at 3:06 PM, [EMAIL PROTECTED] wrote: Brandon Mintern [EMAIL PROTECTED] writes: Call me old school, but that method never takes me more than a few minutes to do. I am also someone in the vim camp. It fires up quickly Unless you are updating a vmappliance built on old (even for 2006) 2006 pkgs to current 2008 pkgs. I know this is a monster thread but making that big a jump is what is being discussed. I can certainly understand and sympathize with that. My first internship and first real exposure to Unix involved updating to a new FreeBSD installation from one that had not been updated in several years. I was mostly responding to Neil Bothwick's implication (even if it was sarcastic) that using vim indicates bad things. -- gentoo-user@lists.gentoo.org mailing list
Re: [gentoo-user] Re: smoothest way to jump from 2006 to 2008
On Thu, 1 May 2008 15:13:27 -0400, Brandon Mintern wrote: I was mostly responding to Neil Bothwick's implication (even if it was sarcastic) that using vim indicates bad things. Implication? I obviously wasn't clear enough ;-) -- Neil Bothwick Just got a new car for my wifeGreat Trade! signature.asc Description: PGP signature
Re: [gentoo-user] Re: smoothest way to jump from 2006 to 2008
On Thursday 01 May 2008, Brandon Mintern wrote: I was mostly responding to Neil Bothwick's implication (even if it was sarcastic) that using vim indicates bad things. Hey Neil, Looks like we caught one - a big fish this time :-) :-) -- Alan McKinnon alan dot mckinnon at gmail dot com -- gentoo-user@lists.gentoo.org mailing list
Re: [gentoo-user] Re: smoothest way to jump from 2006 to 2008
On Thursday 01 May 2008, Neil Bothwick wrote: On Thu, 1 May 2008 15:13:27 -0400, Brandon Mintern wrote: I was mostly responding to Neil Bothwick's implication (even if it was sarcastic) that using vim indicates bad things. Implication? I obviously wasn't clear enough ;-) Hey, watch it there bonzo, or I'll have to send Dirk Gently's friend Noddy and his lads around for a wee chat :-) :-) -- Alan McKinnon alan dot mckinnon at gmail dot com -- gentoo-user@lists.gentoo.org mailing list
Re: [gentoo-user] Re: smoothest way to jump from 2006 to 2008
On Thursday 01 May 2008, Neil Bothwick wrote: On Thu, 1 May 2008 15:13:27 -0400, Brandon Mintern wrote: I was mostly responding to Neil Bothwick's implication (even if it was sarcastic) that using vim indicates bad things. Implication? I obviously wasn't clear enough ;-) You were very clear. Some time ago, I offered you a Hansa Draught for some help if you ever happened to come to Windhoek. That offer is void now, and I'll drink the pint myself. Traitor! Uwe -- Ignorance killed the cat, sir, curiosity was framed! -- gentoo-user@lists.gentoo.org mailing list
Re: [gentoo-user] Re: smoothest way to jump from 2006 to 2008
On Thu, 1 May 2008 22:41:23 +0100, Uwe Thiem wrote: Implication? I obviously wasn't clear enough ;-) You were very clear. Some time ago, I offered you a Hansa Draught for some help if you ever happened to come to Windhoek. That offer is void now, and I'll drink the pint myself. Traitor! Oh, vi! I thought Alan was on about something else. I LOVE vi!!! -- Neil Bothwick EASY TO INSTALL = Difficult to install, but instruction manual has pictures. signature.asc Description: PGP signature
Re: [gentoo-user] Re: smoothest way to jump from 2006 to 2008
On Wednesday 30 April 2008, [EMAIL PROTECTED] wrote:
Alan McKinnon [EMAIL PROTECTED] writes:
I'd like to jump an 2006 install up to 2008. I've never made that
big an update without a fresh install.
There's no such thing as a 2006 install. What does exist, is the
collection of packages that were on the LiveCDs released in 2006.
It's
symantics...
Here is the story behind my phraseology... I've tried 4-5 times to
build a vmware gentoo install hosted on vista home premium. Using
2008 install media.
I cannot seem to get past a kernel panic that appears to be expecting
an intramfs (You may remember this from a previous thread) After
hand rolling 3 different kernels and trying genkernal all its all
ended in the same kernel panic.
Yes, I remember it well.
So tryin to slip in the back way I found a gentoo vmware
appliance online .. downloaded and fired it up on vista. It boots
and runs with no problems. Only thing is it is built on 2006 package
set.
What kernel does it use? I would rather update that vm's kernel to the
latest available and see if that boots properly. If so, move forward
and update the rest of the system. I reason that kernel panics on boot
concern only the boot loader and kernel and have nothing to do with
user-space, so you should concentrate on the more relevant bits.
I also think it's time to bring out the big guns with all the data.
Could you reply and attach the vmware config files for the non-working
vm, plus the .config file for that kernel in the vm?
I strongly suspect you have a simple incompatibility between the
hardware vmware provides that vm and how the kernel is configured.
One blockage that looks to be particularly troublesome is:
Portage blocking bash
bash blocking portage.
That's a nasty one and you quite rightly don't want to unmerge portage
or bash. The trick is to do it in several stages:
First look at portage:
[EMAIL PROTECTED] /var $ eix -e portage
[I] sys-apps/portage
Available versions: 2.0.51.22-r3 2.1.1-r2 2.1.4.4 (~)2.1.5_rc5 (~)
2.1.5_rc6 [M](~)2.2_pre3 [M](~)2.2_pre5 {build doc elibc_FreeBSD
elibc_glibc elibc_uclibc epydoc linguas_pl selinux userland_Darwin
userland_GNU}
Installed versions: 2.1.5_rc6(10:57:02 PM 04/24/2008)
(doc -build -epydoc -linguas_pl -selinux)
[EMAIL PROTECTED] /var/portage/sys-apps/portage $ grep bash *ebuild
portage-2.0.51.22-r3.ebuild:RDEPEND=!build? ( =sys-apps/sed-4.0.5
dev-python/python-fchksum =dev-lang/python-2.2.1 sys-apps/debianutils
=app-shells/bash-2.05a ) !x86-fbsd? ( !mips? (
=sys-apps/sandbox-1.2.17 ) ) selinux? (
=dev-python/python-selinux-2.15 )
portage-2.1.1-r2.ebuild:!userland_Darwin? (
=app-shells/bash-3.0 ) )
portage-2.1.4.4.ebuild: =app-shells/bash-3.2_p17 )
portage-2.1.5_rc5.ebuild: =app-shells/bash-3.2_p17 )
portage-2.1.5_rc6.ebuild: =app-shells/bash-3.2_p17 )
portage-2.2_pre3.ebuild:=app-shells/bash-3.2_p17
portage-2.2_pre5.ebuild:=app-shells/bash-3.2_p17
And look at bash:
[EMAIL PROTECTED] ~ $ eix -e bash
[I] app-shells/bash
Available versions: 2.05b-r11 3.0-r12 (~)3.0-r13 (~)3.0-r14
3.1_p17 3.2_p17-r1 (~)3.2_p33 {afs bashlogger build minimal nls plugins
unicode vanilla}
Installed versions: 3.2_p33(10:01:41 PM 01/05/2008)
(-afs -bashlogger -nls -plugins -vanilla)
[EMAIL PROTECTED] /var/portage/app-shells/bash $ grep portage *ebuild
bash-3.2_p33.ebuild:!sys-apps/portage-2.1.4_rc1
Recent portages want any bash =bash-3.2_p17, but you can't use the most
recent bash (3.2_p33) as that blocks the portage you have. So, do it
manually:
1. emerge --nodeps bash-3.2_p17-r1
2. emerge --nodeps portage
3. emerge bash
4. emerge world
The --nodeps is there to stop portage updating other stuff that
conflicts with what you are trying to do.
Once this is done, you will still have the python/python-updater blocker
to deal with, but I seem to recall posting on that earlier
--
Alan McKinnon
alan dot mckinnon at gmail dot com
--
gentoo-user@lists.gentoo.org mailing list
[gentoo-user] Re: smoothest way to jump from 2006 to 2008
Alan McKinnon [EMAIL PROTECTED] writes: [...] I cannot seem to get past a kernel panic that appears to be expecting an intramfs (You may remember this from a previous thread) After hand rolling 3 different kernels and trying genkernal all its all ended in the same kernel panic. Yes, I remember it well. So tryin to slip in the back way I found a gentoo vmware appliance online .. downloaded and fired it up on vista. It boots and runs with no problems. Only thing is it is built on 2006 package set. What kernel does it use? I would rather update that vm's kernel to the latest available and see if that boots properly. If so, move forward and update the rest of the system. I reason that kernel panics on boot concern only the boot loader and kernel and have nothing to do with user-space, so you should concentrate on the more relevant bits. 2.6.16-r13 and yes as usual you've cut right to a good way to get this resolved. Wish I had thought of it. I think I'll try to stop in the middle here and try that very thing. I'm in the middle of update world dealing with blockages... I got past the bash portage blocker by doing the no-no emerge -vC bash... Then hand building bash and inserting a package.provided listing a higher version of bash. After symlinking /usr/local/bin/bash to /bin/sh and /bin/bash the portage tools seem to work ok. Instead of the much smarter way you laid out. I did try the --nodeps route you suggest but probably not in the smarter order you laid out. I've lost track now what all I tried I probably could have just did the package-provided without emerge -vC bash ... another dumb, with possible later repercussions, move. But now that blockage is solved... I'm getting a failure in the dependancy gpm when I try to emerge -vu portage. I've included that failure message at the end. Right now I think I'll try with --nodeps and if portage installs ok... I'll try emerging and building a more recent kernel and see if is still boots. I also think it's time to bring out the big guns with all the data. Could you reply and attach the vmware config files for the non-working vm, plus the .config file for that kernel in the vm? I strongly suspect you have a simple incompatibility between the hardware vmware provides that vm and how the kernel is configured. Er... it may be a bit late for that...since I trashed the 3 gentoo based installs I had been working with. I discovered ubuntu 8.0.4 install media produced a working vmware install of ubuntu. But only after ditching their desktop install that makes things much harder by making it nearly impossible to get to a root shell from the gui mess. They have it setup where you cannot create a root login and must rely on sudo... in my case sudo would fail with a relolution of host problem before I could run any commands. Rather than try to wade thru that mess.I found there 8.0.4-server install and was able to get a working install I've now built up with apt-get. The ubuntu kernel that is working is their: 2,6,24-server kernel. I have it operational now but I really miss portage... thats why I tried sneaking in the back way with a gentoo appliance. [...] snipped outline of method to resolve blockage bash = portage Thanks for the nifty walk thru.. At this moment in the current gentoo install, I'm going to see if using --nodeps will get portage updated then try a newer kernel and see if it still boots. emerge -vu portage ends with this faiure while emerging dependancies: unpack gpm-1.20.3.tar.lzma: file format not recognized. Ignoring. /usr/portage/sys-libs/gpm/gpm-1.20.3.ebuild: line 24: cd: /var/tmp/portage/gpm-1.20.3/work/gpm-1.20.3: No such file or directory * Applying gpm-1.20.3-no-emacs-dir.patch ... * Failed Patch: gpm-1.20.3-no-emacs-dir.patch ! * ( /usr/portage/sys-libs/gpm/files/gpm-1.20.3-no-emacs-dir.patch ) * * Include in your bugreport the contents of: * * /var/tmp/portage/gpm-1.20.3/temp/gpm-1.20.3-no-emacs-dir.patch-16246.out !!! ERROR: sys-libs/gpm-1.20.3 failed. Call stack: ebuild.sh, line 1539: Called dyn_unpack ebuild.sh, line 711: Called src_unpack gpm-1.20.3.ebuild, line 25: Called epatch '/usr/portage/sys-libs/gpm/files/gpm-1.20.3-no-emacs-dir.patch' eutils.eclass, line 324: Called die -- gentoo-user@lists.gentoo.org mailing list
Re: [gentoo-user] Re: smoothest way to jump from 2006 to 2008
On Wednesday 30 April 2008, [EMAIL PROTECTED] wrote: Seems like you got to a working state - unorthodox method, but it does seem to have worked :-) snip But now that blockage is solved... I'm getting a failure in the dependancy gpm when I try to emerge -vu portage. I've included that failure message at the end. Right now I think I'll try with --nodeps and if portage installs ok... I'll try emerging and building a more recent kernel and see if is still boots. emerge -vu portage ends with this faiure while emerging dependancies: unpack gpm-1.20.3.tar.lzma: file format not recognized. Ignoring. There was a post earlier today about this very thing. Check today's inbox for more details. Apparently you need a very recent portage to use this feature, so I would suggest you fully resolve the bash/portage thing first before trying to emerge world -- Alan McKinnon alan dot mckinnon at gmail dot com -- gentoo-user@lists.gentoo.org mailing list
Re: [gentoo-user] Re: smoothest way to jump from 2006 to 2008
On Wed, Apr 30, 2008 at 7:39 AM, [EMAIL PROTECTED] wrote: I'm going to try a saner approach to getting it updated. I thought the first move would be to set the make.profile to 2007 then emerge --sync Before changing the profile are you clean with emerge -pvDuN --with-bdeps=y world? (Or at a minimum 'system'). Personally I want to be careful about changing profiles with my system not totally up to date. I'm also running eix-test-obsolete to make sure I'm not masking something I should really be using. So, just sitting here thinking, this is likely overkill but I think safe: eix-sync eix-test-obsolete emerge -pvDuN --with-bdeps=y world emerge -p --depclean emerge -pvDuN --with-bdeps=y world revdep-rebuild -p eix-test-obsolete Assuming everything is totally clean, or at least understood, now I change the profile and repeat the above tasks. Good luck, Mark -- gentoo-user@lists.gentoo.org mailing list
Re: [gentoo-user] Re: smoothest way to jump from 2006 to 2008
On Wed, 30 Apr 2008 07:49:25 -0700, Mark Knecht wrote: eix-sync eix-test-obsolete emerge -pvDuN --with-bdeps=y world emerge -p --depclean emerge -pvDuN --with-bdeps=y world revdep-rebuild -p eix-test-obsolete I'd run the first eix-test-obsolete after emerge --depclean since adding. removing or updating packages can change the relevance of entries in /etc/portage/package.* -- Neil Bothwick A pessimist complains about the noise when opportunity knocks. signature.asc Description: PGP signature
Re: [gentoo-user] Re: smoothest way to jump from 2006 to 2008
On Wed, Apr 30, 2008 at 8:09 AM, Neil Bothwick [EMAIL PROTECTED] wrote: On Wed, 30 Apr 2008 07:49:25 -0700, Mark Knecht wrote: eix-sync eix-test-obsolete emerge -pvDuN --with-bdeps=y world emerge -p --depclean emerge -pvDuN --with-bdeps=y world revdep-rebuild -p eix-test-obsolete I'd run the first eix-test-obsolete after emerge --depclean since adding. removing or updating packages can change the relevance of entries in /etc/portage/package.* -- Neil Bothwick Agreed Neil. Actually I'm pretty much running that things after every step. It only takes about 20 seconds to run and is a nice early warning system for things to come. What I'd really like would be for eix-test-obsolete to have an option to run against the remote rsync server *before* I run eix-sync so that I can see what my issues will be after the sync operation removes things. That would give me a nice heads up that I need to archive something before I get sent to the CVS attic. Thanks, Mark -- gentoo-user@lists.gentoo.org mailing list
Re: [gentoo-user] Re: smoothest way to jump from 2006 to 2008
On Wed, Apr 30, 2008 at 8:29 AM, Mark Knecht [EMAIL PROTECTED] wrote: On Wed, Apr 30, 2008 at 8:09 AM, Neil Bothwick [EMAIL PROTECTED] wrote: On Wed, 30 Apr 2008 07:49:25 -0700, Mark Knecht wrote: eix-sync eix-test-obsolete emerge -pvDuN --with-bdeps=y world emerge -p --depclean emerge -pvDuN --with-bdeps=y world revdep-rebuild -p eix-test-obsolete I'd run the first eix-test-obsolete after emerge --depclean since adding. removing or updating packages can change the relevance of entries in /etc/portage/package.* -- Neil Bothwick Agreed Neil. Actually I'm pretty much running that things after every step. It only takes about 20 seconds to run and is a nice early warning system for things to come. What I'd really like would be for eix-test-obsolete to have an option to run against the remote rsync server *before* I run eix-sync so that I can see what my issues will be after the sync operation removes things. That would give me a nice heads up that I need to archive something before I get sent to the CVS attic. Thanks, Mark Actually, I'm having a bit of trouble with my AMD64 machine not performing very well as a MythTV frontend machine since the upgrade to mythtv-0.21. Prior to 0.21 it worked great. It also works great when booted into Windows running the MythTV Player for Windows and the NetFlix streaming service so I'm confident there isn't anything wrong with the hardware. I noticed, following through the above flow, that this one machine was still using the 2006.0 desktop profile whereas my other machines that work fine (albeit very different hardware) are all 2007.0 desktop machines. That gave me an opportunity to put some of this thread to work. 1) The machine was completely clean with the flow above using 2006.0. 2) I switched to 2007.0 desktop and emerge -DuN world wanted to rebuild 39 packages. 3) I switched back to 2006.0 and the machine is clean again, as it should be. I just wanted the sanity check. 4) I switched into 2007.0 desktop and did an emerge -pvDuN system (instead of world) and saw that it wanted to rebuild 9 packages so I've decided to do that first, then sync again, and then look at emerge -DuN world again to see what, if anything, changes. No idea if this will fix my machine's MythTV problems but it's probably a good thing to do anyway. Cheers, Mark -- gentoo-user@lists.gentoo.org mailing list
[gentoo-user] Re: smoothest way to jump from 2006 to 2008
Mark Knecht [EMAIL PROTECTED] writes: eix-sync eix-test-obsolete emerge -pvDuN --with-bdeps=y world emerge -p --depclean -p --depclean tells me: !!! You have no system list. What does that mean? Also: emerge -pvDuN --with-bdeps=y world neither man emerge nor man portage show any hits on bdeps. Where can I find out what it does? Here it just errors out like this: !!! Error: --with-bdeps=y is an invalid option. -- gentoo-user@lists.gentoo.org mailing list
Re: [gentoo-user] Re: smoothest way to jump from 2006 to 2008
On Wed, Apr 30, 2008 at 11:00 AM, [EMAIL PROTECTED] wrote: Mark Knecht [EMAIL PROTECTED] writes: eix-sync eix-test-obsolete emerge -pvDuN --with-bdeps=y world emerge -p --depclean emerge -pvDuN --with-bdeps=y world revdep-rebuild -p eix-test-obsolete Assuming everything is totally clean, or at least understood, now I change the profile and repeat the above tasks. What does it mean or what is called for if eix-test-obsolete returns hundreds of packages? Installed packages with a version not in the database (or masked): [U] app-admin/eselect ([EMAIL PROTECTED]/21/06 - (~*)1.0.11-r1): Modular -config replacement utility In general it means that the version of the package (eselect in this case) has been removed from the portage tree by the folks who maintain portage. That means that should you want or need that specific version you cannot get it from portage doing something like this: emerge -pv =app-admin/eselect-1.0.2 The general path I'm taking at this point is that all of these packages are out of date and should be updated. In general they will be with the emerge -DuN world operation. However when I run into a list that is that large I try to handle it in pieces: emerge -DuN system emerge -DuN gnome emerge -DuN mythTV etc., where I work through the list in a few steps instead of just one pass. (The above are just examples, not suggestions.) The issue here, and you're free to assume otherwise, is that I personally didn't want to switch profiles until I was completely clean as I wouldn't understand what changes were taking place due to the profile change and which were because I wasn't up to date anyway. - Mark -- gentoo-user@lists.gentoo.org mailing list
Re: [gentoo-user] Re: smoothest way to jump from 2006 to 2008
On Wednesday 30 April 2008, [EMAIL PROTECTED] wrote: Mark Knecht [EMAIL PROTECTED] writes: eix-sync eix-test-obsolete emerge -pvDuN --with-bdeps=y world emerge -p --depclean emerge -pvDuN --with-bdeps=y world revdep-rebuild -p eix-test-obsolete Assuming everything is totally clean, or at least understood, now I change the profile and repeat the above tasks. What does it mean or what is called for if eix-test-obsolete returns hundreds of packages? It means you need to trim the contents of /etc/portage/package* :-) You probably have hundreds of USE flags set in package.use that are already in make.conf, or entries in various files for packages that are no longer installed, just to give 2 random examples. eix-test-obsolete groups it's output with headings above each group. The heading tells you enough to decide if you can trim stuff or not Installed packages with a version not in the database (or masked): [U] app-admin/eselect ([EMAIL PROTECTED]/21/06 - (~*)1.0.11-r1): Modular -config replacement utility This one os marked ~x86 in the tree. Possibly you explicitly merged it with an entry in package.keywords, then removed that entry later after merging the package. -- Alan McKinnon alan dot mckinnon at gmail dot com -- gentoo-user@lists.gentoo.org mailing list
[gentoo-user] Re: smoothest way to jump from 2006 to 2008
Mark Knecht [EMAIL PROTECTED] writes: On Wed, Apr 30, 2008 at 11:08 AM, [EMAIL PROTECTED] wrote: Mark Knecht [EMAIL PROTECTED] writes: eix-sync eix-test-obsolete emerge -pvDuN --with-bdeps=y world emerge -p --depclean -p --depclean tells me: !!! You have no system list. What does that mean? That someone above my pay grade will have to answer your question. Also: emerge -pvDuN --with-bdeps=y world neither man emerge nor man portage show any hits on bdeps. Where can I find out what it does? Here it just errors out like this: !!! Error: --with-bdeps=y is an invalid option. What version of portage are you running? root # eix -Ic portage [U] sys-apps/portage ([EMAIL PROTECTED]/20/06 - (~)2.1.5_rc6) As I mentioned early on in this thread. I'm working with an out of date 2006 based vmappliance, due to inablity to get a bootable vmware install of gentoo-2008 based to work. I have a bootable vmware appliance of gentoo (hosted on vista) that works so am trying to update it. [...] QUOTE --with-bdeps y | n Thanks -- gentoo-user@lists.gentoo.org mailing list
[gentoo-user] Re: smoothest way to jump from 2006 to 2008
Mark Knecht [EMAIL PROTECTED] writes: eix-sync eix-test-obsolete emerge -pvDuN --with-bdeps=y world emerge -p --depclean emerge -pvDuN --with-bdeps=y world revdep-rebuild -p eix-test-obsolete Assuming everything is totally clean, or at least understood, now I change the profile and repeat the above tasks. What does it mean or what is called for if eix-test-obsolete returns hundreds of packages? Installed packages with a version not in the database (or masked): [U] app-admin/eselect ([EMAIL PROTECTED]/21/06 - (~*)1.0.11-r1): Modular -config replacement utility [...] [U] x11-themes/gnome-icon-theme ([EMAIL PROTECTED]/24/06 - (~*)2.22.0): GNOME 2 default icon themes [D] x11-themes/hicolor-icon-theme ([EMAIL PROTECTED]/24/06 - ??): Fallback theme for the freedesktop icon theme specifica\ tion [U] x11-wm/twm ([EMAIL PROTECTED]/21/06 - (~*)1.0.4): X.Org twm application *** Found 348 matches. [emphasis addded -ed] -- gentoo-user@lists.gentoo.org mailing list
Re: [gentoo-user] Re: smoothest way to jump from 2006 to 2008
On Wednesday 30 April 2008, [EMAIL PROTECTED] wrote: Mark Knecht [EMAIL PROTECTED] writes: eix-sync eix-test-obsolete emerge -pvDuN --with-bdeps=y world emerge -p --depclean -p --depclean tells me: !!! You have no system list. What does that mean? Trouble, big trouble. Does /etc/make.profile point to an actual existing profile? That's the only thing I can think of that takes system away. It's defined in the various files named packages in the profile directories. Also: emerge -pvDuN --with-bdeps=y world neither man emerge nor man portage show any hits on bdeps. Where can I find out what it does? 'man emerge' line 350 9for version 2.1.5_rc6 Here it just errors out like this: !!! Error: --with-bdeps=y is an invalid option. it's --with-bdeps y (no minus/dash) -- Alan McKinnon alan dot mckinnon at gmail dot com -- gentoo-user@lists.gentoo.org mailing list
Re: [gentoo-user] Re: smoothest way to jump from 2006 to 2008
On Wed, Apr 30, 2008 at 11:36 AM, Alan McKinnon [EMAIL PROTECTED] wrote: SNIP !!! Error: --with-bdeps=y is an invalid option. it's --with-bdeps y (no minus/dash) So right you are Alan. Thanks. Sorry to reader. cheers, Mark -- gentoo-user@lists.gentoo.org mailing list
[gentoo-user] Re: smoothest way to jump from 2006 to 2008
Alan McKinnon [EMAIL PROTECTED] writes: [...] -p --depclean tells me: !!! You have no system list. What does that mean? Trouble, big trouble. Does /etc/make.profile point to an actual existing profile? That's the only thing I can think of that takes system away. It's defined in the various files named packages in the profile directories. Yes it pointed to 2007 and was set with eselect profile set But anyway emerge tells you immediately when there is no profile. But there is good news... and it harks back to your second post on this monster thread: [...] Harry wrote: One blockage that looks to be particularly troublesome is: Portage blocking bash bash blocking portage. Alan replied: That's a nasty one and you quite rightly don't want to unmerge portage or bash. The trick is to do it in several stages: First look at portage: [...] Recent portages want any bash =bash-3.2_p17, but you can't use the most recent bash (3.2_p33) as that blocks the portage you have. So, do it manually: 1. emerge --nodeps bash-3.2_p17-r1 2. emerge --nodeps portage 3. emerge bash 4. emerge world The --nodeps is there to stop portage updating other stuff that conflicts with what you are trying to do. Once this is done, you will still have the python/python-updater blocker to deal with, but I seem to recall posting on that earlier That list error occurred prior to getting your above suggestion done. At that point emerge -vup system or really an emerge -u showed a about 12-15 blockages including portage and bash Once I followed the recipe above all that disappeared along with the list error. I think that older version of portage was really what was causing most of the trouble. At this point I have profile set to 2008-server and have completed an eix-sync with that profile active. I've now also trimmed the use list in make.conf down removing any x related ones and only using a few I know I want and a few left from the prior list that looked harmless. Also added a few `-' ones to be sure. (I hate maildir for example.) grep USE /etc/make.conf: USE= -arts -X -xorg -maildir mbox emacs logrotate readline samba smb usb zlib Some may be redundant there.. but I doubt any will get me in trouble. I've test run emerge -vuDNp world and see no blockages. I'd like to get rid of a bunch of X stuff but not really sure how best to go at it. I really don't want to set thru the grinding of kde and really don't want X at all on this vmware. Would it work to that end if I edit the world file removing x related apps before running emerge -vuDNp world and with the new use flags in place. Or will that leave herds of stuff on board? -- gentoo-user@lists.gentoo.org mailing list
Re: [gentoo-user] Re: smoothest way to jump from 2006 to 2008
On Wednesday 30 April 2008, [EMAIL PROTECTED] wrote: Once I followed the recipe above all that disappeared along with the list error. I think that older version of portage was really what was causing most of the trouble. Makes sense. Portage is rather fond of finding the kind of files and directories it expects to find grin funny that big grin At this point I have profile set to 2008-server and have completed an eix-sync with that profile active. I've now also trimmed the use list in make.conf down removing any x related ones and only using a few I know I want and a few left from the prior list that looked harmless. Also added a few `-' ones to be sure. (I hate maildir for example.) grep USE /etc/make.conf: USE= -arts -X -xorg -maildir mbox emacs logrotate readline samba smb usb zlib Some may be redundant there.. but I doubt any will get me in trouble. Looks fine from here I've test run emerge -vuDNp world and see no blockages. I'd like to get rid of a bunch of X stuff but not really sure how best to go at it. I really don't want to set thru the grinding of kde and really don't want X at all on this vmware. Would it work to that end if I edit the world file removing x related apps before running emerge -vuDNp world and with the new use flags in place. Or will that leave herds of stuff on board? There are two ways, I've tried them both. One is long, boring and relatively easy [1] the other is potentially shorter, exciting, a test of your skill and proves you have TheRightStuff [2] much as editing main.cf by hand proves the same... [1] emerge -avuND world, resolve blocks, run to completion. emerge -av --depclean, make sure nothing desired will go away. If so, emerge -n package you want to keep Rinse, repeat emerge -avuND world to check nothing more to be done revdep-rebuild -p -i as a sanity check [2] vi /var/lib/portage/world Edit at will with sense of abandon vi /etc/make.conf Edit where appropriate vi /etc/portage/* Fearlessly edit throwing caution to the winds loop_entry: emerge -avuND world Resolve blockers emerge --resume --skipfirst (bonus points if you correctly predict which packages will fail to build) emerge -av --depclean goto loop_entry Having recently been through [2] I recommend you do [1] instead. In an effort to avoid remerging all of gnome (which I wanted to remove anyway, and takes about 8 hours to build) I caused myself to battle with emerge --resume --skipfirst over the course of 4 days. Note that the machine was mostly unusable for those 4 days. Go figure. -- Alan McKinnon alan dot mckinnon at gmail dot com -- gentoo-user@lists.gentoo.org mailing list
[gentoo-user] Re: smoothest way to jump from 2006 to 2008
Alan McKinnon [EMAIL PROTECTED] writes: [2] vi /var/lib/portage/world What if I emerge -vC all I know I don't want. All kde all gnome all xorg for example, before update world. I was hoping to accomplish much the same thing by editing world. But strangely I see only a few candidates to delete (marked with asterisks): /var/lib/portage/world: app-admin/eselect app-admin/sysklogd app-editors/emacs app-editors/vim app-misc/screen ** app-office/openoffice-bin app-portage/eix app-portage/gentoolkit ** kde-base/kde net-misc/dhcpcd sys-apps/slocate sys-boot/grub sys-devel/gdb sys-kernel/gentoo-sources ** www-client/seamonkey ** x11-drivers/xf86-input-keyboard ** x11-drivers/xf86-input-mouse ** x11-drivers/xf86-video-fbdev ** x11-drivers/xf86-video-vesa ** x11-drivers/xf86-video-vga ** x11-drivers/xf86-video-vmware However eix -I shows a whole bunch of kde packages. Not to mention piles of other stuff not mentioned in `world'. Looks like emerge -vC ???, in conjunction with `2' would seriously cut down the build time even if I didn't weed out everything on the first pass. Or will I likely hit a nightmare of dependency problems if I happen to emerge -vC the wrong library or whatever. I'm willing to chase down equery depends to a degree but not really thoroughly. That would take most of the remainder of my lifetime. -- gentoo-user@lists.gentoo.org mailing list
Re: [gentoo-user] Re: smoothest way to jump from 2006 to 2008
On Wed, Apr 30, 2008 at 3:31 PM, [EMAIL PROTECTED] wrote: Alan McKinnon [EMAIL PROTECTED] writes: [2] vi /var/lib/portage/world What if I emerge -vC all I know I don't want. All kde all gnome all xorg for example, before update world. I was hoping to accomplish much the same thing by editing world. But strangely I see only a few candidates to delete (marked with asterisks): /var/lib/portage/world: app-admin/eselect app-admin/sysklogd app-editors/emacs app-editors/vim app-misc/screen ** app-office/openoffice-bin app-portage/eix app-portage/gentoolkit ** kde-base/kde net-misc/dhcpcd sys-apps/slocate sys-boot/grub sys-devel/gdb sys-kernel/gentoo-sources ** www-client/seamonkey ** x11-drivers/xf86-input-keyboard ** x11-drivers/xf86-input-mouse ** x11-drivers/xf86-video-fbdev ** x11-drivers/xf86-video-vesa ** x11-drivers/xf86-video-vga ** x11-drivers/xf86-video-vmware However eix -I shows a whole bunch of kde packages. Not to mention piles of other stuff not mentioned in `world'. My rule, for the most part, is that nothing is in /var/lib/portage/world that I don't know I need and that I run by hand. for instance here is a MythTV backend server I brought up 3 weeks ago: Sector9 ~ # cat /var/lib/portage/world app-admin/sudo app-admin/syslog-ng app-editors/vim app-misc/screen app-portage/eix app-portage/gentoolkit app-portage/layman dev-db/mysql media-libs/alsa-oss media-sound/alsa-tools media-sound/alsa-utils media-tv/ivtv media-tv/mythtv media-video/nvidia-settings net-misc/ntp net-print/cups net-print/foomatic net-print/gutenprint sys-apps/baselayout sys-apps/hotplug sys-apps/slocate sys-apps/usbutils sys-boot/grub sys-devel/gcc:3.4 sys-devel/gcc:4.1 sys-kernel/gentoo-sources sys-process/vixie-cron x11-base/xorg-x11 x11-drivers/nvidia-drivers Sector9 ~ # It's pretty clean. The nvidia drivers can probably go as it will mainly be headless. The cups/gutenprint are planning for making it a print server pretty soon. I franlky don't remember much about hotplug anymore. Maybe it can go. No packages selected for removal by depclean To see reverse dependencies, use --verbose Packages installed: 334 Packages in world:29 Packages in system: 52 Unique package names: 334 Required packages:334 Number to remove: 0 Sector9 ~ # Anyway, if I have the time I reduce the contents of this file and let emerge --depclean tell me what to clean out. If it's a big list then often I'll do a few by hand, in groups of 5 or so, as copy/paste options. Look at the list, choose 5, emerge -C them, and then do it again. This gives me a little more confidence that I won't emerge -C some system file and cause big problems. One last thing that I've done once or twice when the whole process is complete - I don't think it's mentioned here so far and I don't always do it - but for complete consistency I've gone so far as: emerge -e system emerge -e world and let everything rebuild thus guaranteeing that nothing is left to chance. I usually do that over a weekend or sometime where usage is going to be lower, but as I say I'd only go that far after really big changes or if I was seeing flakey log messages somewhere. Probably a waste of time but it's just a computer and once the list is smaller it's not as slow as you think, especially on a server. Cheers, Mark -- gentoo-user@lists.gentoo.org mailing list
Re: [gentoo-user] Re: smoothest way to jump from 2006 to 2008
On Wed, 30 Apr 2008 22:17:34 +0200, Alan McKinnon wrote: [2] vi /var/lib/portage/world Edit at will with sense of abandon vi /etc/make.conf Edit where appropriate vi /etc/portage/* Fearlessly edit throwing caution to the winds loop_entry: emerge -avuND world Resolve blockers emerge --resume --skipfirst (bonus points if you correctly predict which packages will fail to build) emerge -av --depclean goto loop_entry What a horrible suggestion, not only does it use vi - three time! - but it ends with a goto! -- Neil Bothwick Top Oxymorons Number 5: Twelve-ounce pound cake signature.asc Description: PGP signature
[gentoo-user] Re: smoothest way to jump from 2006 to 2008
Ian Graeme Hilt wrote: Download the 2008 minimal install cd and install. Frankly, updating a 2006 install to 2008 is counter-productive. You'll have a much easier time doing a fresh 2008 install. Why? If he kept his box up-to-date, there are just going to be some slight changes in the default use flags and some virtuals, probably not all that much to re-emerge. That's the beauty of gentoo: Continuous updating making big-bang upgrades unnecessary. Incremental changes are (usually, except when sombody decides to shuffle baselayout all around) small and can be dealt with easily. If you have kept your box up-to-date even on a 2006 profile, you will be current ebuild-version-wise, there's no need to throw that away. This box, for example, has been continuously kept up to date since Gentoo 1.4, before profiles were numbered by year (I think that was in 2004). Not a single reinstall was necessary, or even more productive than a simple update. Of course, I emerge --sync and emerge -auD weekly. -- gentoo-user@lists.gentoo.org mailing list
[gentoo-user] Re: smoothest way to jump from 2006 to 2008
Etaoin Shrdlu wrote: I don't want to comment about the change. However, you could have found out easily all the available profiles by doing eselect profile list: That's a cool eselect module, didn't know about that. I don't change profiles all that often (duh), and end up having to search for the new thingies manually. The automatic list is nice. -- gentoo-user@lists.gentoo.org mailing list
[gentoo-user] Re: smoothest way to jump from 2006 to 2008
Alan McKinnon [EMAIL PROTECTED] writes: I'd like to jump an 2006 install up to 2008. I've never made that big an update without a fresh install. There's no such thing as a 2006 install. What does exist, is the collection of packages that were on the LiveCDs released in 2006. It's symantics... Here is the story behind my phraseology... I've tried 4-5 times to build a vmware gentoo install hosted on vista home premium. Using 2008 install media. I cannot seem to get past a kernel panic that appears to be expecting an intramfs (You may remember this from a previous thread) After hand rolling 3 different kernels and trying genkernal all its all ended in the same kernel panic. So tryin to slip in the back way I found a gentoo vmware appliance online .. downloaded and fired it up on vista. It boots and runs with no problems. Only thing is it is built on 2006 package set. I want to update .. rebuild ... renew ... go forward to ... advance upon .. [or some other preferred terminology] the 2008 pkg set, and not loose the ability to boot the vmware gentoo os on a vista host.. So far I've sinced and upped the profile to 2007 and am now working toward update world... there are a number of packages blocking others so I'm working my way thru those right now. I think I can get to 2007 by hook or by crook but I'm leary that those kernels are going to break the vmware bootability on vista... we'll see. One blockage that looks to be particularly troublesome is: Portage blocking bash bash blocking portage. Usually I've found with blockages... I can unmerge one of the offenders and get on with update world. Unmerging portage is obviously a non-starter and I suspect unmerging bash will have serious repercussions too. I thought I might emerge ksh and make it roots shell then unmerge bash and see if I can proceed with update world, and get bash back further along. -- gentoo-user@lists.gentoo.org mailing list
Re: [gentoo-user] Re: smoothest way to jump from 2006 to 2008
Well I am not sure, but maybe there is a way of adding more space on your appliance, if neccessary. The idea that strikes me, is you could use your current appliance as the Gentoo-life-system and you install a new machine in that free area, chroot to it and work further. Really the classical fresh gentoo install way. You could work and experiment on that new partition without breaking your working appliance. I haven't done this in a vmware environment yet. So it is really just a thought. Cheers. Sandro -- gentoo-user@lists.gentoo.org mailing list
