[gentoo-user] New profile 17: How urgent is the rebuild of world technically?

[gentoo-user] New profile 17: How urgent is the rebuild of world technically?

[tuxic]

Hi,

the instruction of the news item about the swicth to profile 17 says
that one needs to rebuild @world.

Is this technically needed?
Would it be possible to do this on base of the daily updates intead
all in one go?

Background: I simply need my PC more often as it would allow me
to wait (and fix...see previous mails in this mailing list) a
rebuild of world in one go.

Thanks a lot for any help in advance!

Cheers
Meino

Re: [gentoo-user] New profile 17: How urgent is the rebuild of world technically?

[Michael Orlitzky]

On 12/02/2017 09:15 PM, tuxic@posteo.de wrote:
> Hi,
> 
> the instruction of the news item about the swicth to profile 17 says
> that one needs to rebuild @world.
> 
> Is this technically needed?

Yes, unless you were already using a hardened toolchain.


> Would it be possible to do this on base of the daily updates intead
> all in one go?

No, you'll get build failures trying to link new packages against old
libraries.

However, you can delay switching to the new profile for a while.

Re: [gentoo-user] New profile 17: How urgent is the rebuild of world technically?

[Adam Carter]

[-- Attachment #1: Type: text/plain, Size: 419 bytes --]

On Sun, Dec 3, 2017 at 1:30 PM, Michael Orlitzky <mjo@gentoo.org> wrote:

> On 12/02/2017 09:15 PM, tuxic@posteo.de wrote:
> > Hi,
> >
> > the instruction of the news item about the swicth to profile 17 says
> > that one needs to rebuild @world.
> >
> > Is this technically needed?
>
> Yes, unless you were already using a hardened toolchain.
>

Does having the hardened USE flag enabled = having a hardened toolchain?

[-- Attachment #2: Type: text/html, Size: 849 bytes --]

Re: [gentoo-user] New profile 17: How urgent is the rebuild of world technically?

[Michael Orlitzky]

On 12/02/2017 09:32 PM, Adam Carter wrote:
> 
> Does having the hardened USE flag enabled = having a hardened toolchain?

If only it were that simple... what you really need to know is, did you
build everything on your system with PIE enabled?

  * Some packages have "pie" USE flags, and it's only forced-on in the
    hardened profiles. I think that flag may actually have been masked
    in the default profiles?

  * Even if you /built/ a hardened compiler, you can switch it off
    with gcc-config.

  * Your local flags in make.conf can disable some of that stuff, too.

If you were using a hardened *profile*, then chances are that you won't
need to rebuild (unless you switched to a non-hardened compiler on
purpose). Otherwise, I would play it safe and rebuild everything. The
newer GCC probably produces more efficient code anyway, and you will
preempt all of the inevitable problems that no one thought of and that
weren't mentioned in the news item.

Re: [gentoo-user] New profile 17: How urgent is the rebuild of world technically?

[tuxic]

On 12/02 09:30, Michael Orlitzky wrote:
> On 12/02/2017 09:15 PM, tuxic@posteo.de wrote:
> > Hi,
> > 
> > the instruction of the news item about the swicth to profile 17 says
> > that one needs to rebuild @world.
> > 
> > Is this technically needed?
> 
> Yes, unless you were already using a hardened toolchain.
> 
> 
> > Would it be possible to do this on base of the daily updates intead
> > all in one go?
> 
> No, you'll get build failures trying to link new packages against old
> libraries.
> 
> However, you can delay switching to the new profile for a while.
> 

Delaying would not solve the problem as it is...

If the compilation will fail at a certain point (and it will fail,
since this is a complete new thing) -- would it be possible to resume
even some tweaks, hacks and patches (even certain recompilations)
would be needed in between?

Can I stop a running emerge @world and resume later?

How does a restarted emerge @world recognizes packages, which are
already compiled according to the new standard?

Cheers
Meino

Re: [gentoo-user] New profile 17: How urgent is the rebuild of world technically?

[Heiko Baums]

Am Sun, 3 Dec 2017 04:26:55 +0100
schrieb tuxic@posteo.de:

> If the compilation will fail at a certain point (and it will fail,
> since this is a complete new thing) -- would it be possible to resume
> even some tweaks, hacks and patches (even certain recompilations)
> would be needed in between?

Just run `emerge -e --keep-going y @world`.

> Can I stop a running emerge @world and resume later?

Maybe with `emerge --resume`. But I don't know if interrupting this
would cause some problems in this particular case.

> How does a restarted emerge @world recognizes packages, which are
> already compiled according to the new standard?

It simply creates a list of the packages to be installed as usual and
knows which of them are already installed and which are not. Then it
recalculates the dependency tree as usual.

Heiko

Re: [gentoo-user] New profile 17: How urgent is the rebuild of world technically?

[tuxic]

On 12/03 04:35, Heiko Baums wrote:
> Am Sun, 3 Dec 2017 04:26:55 +0100
> schrieb tuxic@posteo.de:
> 
> > If the compilation will fail at a certain point (and it will fail,
> > since this is a complete new thing) -- would it be possible to resume
> > even some tweaks, hacks and patches (even certain recompilations)
> > would be needed in between?
> 
> Just run `emerge -e --keep-going y @world`.
> 
> > Can I stop a running emerge @world and resume later?
> 
> Maybe with `emerge --resume`. But I don't know if interrupting this
> would cause some problems in this particular case.
> 
> > How does a restarted emerge @world recognizes packages, which are
> > already compiled according to the new standard?
> 
> It simply creates a list of the packages to be installed as usual and
> knows which of them are already installed and which are not. Then it
> recalculates the dependency tree as usual.
> 
> Heiko
> 

Hi Heiko,

...sorry my question was unclear.

Suppose one would do an emerge @world...and then BOOOM! a powerfailyre
would stop the whole thing. Further suppose the filesystem, the
hardware and anything has survived luckily -- only emerge @world needs
to be restarted.
And one does NOT an emerge --resume but an emerge @world.
In this particular case...how does emerge knows from the previous
emerge @world what packages has been recompiled already and are "PIE"?

How can I check, whether a binary is "PIE"-conform ("pie-conform" is
a freaky funny language hack :) ;) )  ?

Cheers
Meino

Re: [gentoo-user] New profile 17: How urgent is the rebuild of world technically?

[Heiko Baums]

Am Sun, 3 Dec 2017 04:26:55 +0100
schrieb tuxic@posteo.de:

> Delaying would not solve the problem as it is...

Btw., if you're using the testing tree (e.g. ~amd64) you should do a gcc
upgrade instead of rebuilding gcc as the news item says.

Unfortunately the news item explicitly mentions gcc-6.4.0. After
rebuilding it as the news item says I found out that gcc-7.2.0 is
already in the portage tree.

So more than an hour wasted.

Heiko

Re: [gentoo-user] New profile 17: How urgent is the rebuild of world technically?

[Heiko Baums]

Am Sun, 3 Dec 2017 04:45:59 +0100
schrieb tuxic@posteo.de:

> Suppose one would do an emerge @world...and then BOOOM! a powerfailyre
> would stop the whole thing.

In such a case you should consider buying a UPS.

Can't you do this over night or a weekend? And how often do you have a
power failure?

> Further suppose the filesystem, the
> hardware and anything has survived luckily -- only emerge @world needs
> to be restarted.
> And one does NOT an emerge --resume but an emerge @world.

If I'm not mistaken in this case an `emerge --resume` should be
the right thing. It usually resumes the last interrupted emerge command.

That is if you run `emerge -e @world` press Ctrl-C `emerge --resume`
should resume this previous started `emerge -e @world` including the
package which was built when it was interrupted. Ctrl-C is principally
the same as a power failure.

If this fails for some reason then you'd have to rerun the whole
`emerge -e @world` I guess.

> In this particular case...how does emerge knows from the previous
> emerge @world what packages has been recompiled already and are "PIE"?

Like I said before. emerge always calculates the dependency tree, which
is a lot faster in case of `emerge -e @world` than in case of `emerge
-uDN @world`. And then it knows which packages have already been
installed and which are not.

That said I haven't run an `emerge -e @world` before. So I'm actually
not sure if this works the same way as with an `emerge -uDN @world`.

Heiko

Re: [gentoo-user] New profile 17: How urgent is the rebuild of world technically?

[Adam Carter]

[-- Attachment #1: Type: text/plain, Size: 361 bytes --]

> How can I check, whether a binary is "PIE"-conform ("pie-conform" is
> a freaky funny language hack :) ;) )  ?
>

emerge hardening-check

# hardening-check /usr/bin/cpuid2cpuflags
/usr/bin/cpuid2cpuflags:
 Position Independent Executable: yes
 Stack protected: yes
 Fortify Source functions: yes
 Read-only relocations: yes
 Immediate binding: no, not found!

[-- Attachment #2: Type: text/html, Size: 708 bytes --]

Re: [gentoo-user] New profile 17: How urgent is the rebuild of world technically?

[Peter Humphrey]

On Sunday, 3 December 2017 04:15:25 GMT Heiko Baums wrote:

> Like I said before. emerge always calculates the dependency tree, which
> is a lot faster in case of `emerge -e @world` than in case of `emerge
> -uDN @world`. And then it knows which packages have already been
> installed and which are not.
> 
> That said I haven't run an `emerge -e @world` before. So I'm actually
> not sure if this works the same way as with an `emerge -uDN @world`.

Nope. Empty-tree means empty-tree. That is, whenever you emerge -e world, 
you start from the beginning every time, regardless of anything you were 
doing just before that.

-- 
Regards,
Peter.

Re: [gentoo-user] New profile 17: How urgent is the rebuild of world technically?

[Neil Bothwick]

[-- Attachment #1: Type: text/plain, Size: 1030 bytes --]

On Sun, 3 Dec 2017 04:45:59 +0100, tuxic@posteo.de wrote:

> Suppose one would do an emerge @world...and then BOOOM! a powerfailyre
> would stop the whole thing. Further suppose the filesystem, the
> hardware and anything has survived luckily -- only emerge @world needs
> to be restarted.
> And one does NOT an emerge --resume but an emerge @world.
> In this particular case...how does emerge knows from the previous
> emerge @world what packages has been recompiled already and are "PIE"?

Of course it doesn't, it only does what you tell it to do. If you tell it
to resume where it left off, it will do that. If you tell it to rebuild
everything, with emerge -e @world, it will do that.

Portage, like any other program, does not know what you want it to do,
only what you tell it to do.

If you want to know which packages have already been rebuilt, use qlop or
check the timestamp of /var/db/pkg/cat/name-ver/environment.bz2.


-- 
Neil Bothwick

If at first you don't succeed, call in an airstrike.

[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 833 bytes --]

Re: [gentoo-user] New profile 17: How urgent is the rebuild of world technically?

[Heiko Baums]

Am Sun, 03 Dec 2017 09:53:21 +0000
schrieb Peter Humphrey <peter@prh.myzen.co.uk>:

> On Sunday, 3 December 2017 04:15:25 GMT Heiko Baums wrote:
> 
> > Like I said before. emerge always calculates the dependency tree,
> > which is a lot faster in case of `emerge -e @world` than in case of
> > `emerge -uDN @world`. And then it knows which packages have already
> > been installed and which are not.
> > 
> > That said I haven't run an `emerge -e @world` before. So I'm
> > actually not sure if this works the same way as with an `emerge
> > -uDN @world`.  
> 
> Nope. Empty-tree means empty-tree. That is, whenever you emerge -e
> world, you start from the beginning every time, regardless of
> anything you were doing just before that.

Actually I was talking about the behavior of `emerge --resume` in the
case of `emerge -e @world` compared to `emerge -uDN @world`. Sorry, if
this was unclear.

Heiko

Re: [gentoo-user] New profile 17: How urgent is the rebuild of world technically?

[Dale]

Heiko Baums wrote:
> Am Sun, 03 Dec 2017 09:53:21 +0000
> schrieb Peter Humphrey <peter@prh.myzen.co.uk>:
>
>> On Sunday, 3 December 2017 04:15:25 GMT Heiko Baums wrote:
>>
>>> Like I said before. emerge always calculates the dependency tree,
>>> which is a lot faster in case of `emerge -e @world` than in case of
>>> `emerge -uDN @world`. And then it knows which packages have already
>>> been installed and which are not.
>>>
>>> That said I haven't run an `emerge -e @world` before. So I'm
>>> actually not sure if this works the same way as with an `emerge
>>> -uDN @world`.  
>> Nope. Empty-tree means empty-tree. That is, whenever you emerge -e
>> world, you start from the beginning every time, regardless of
>> anything you were doing just before that.
> Actually I was talking about the behavior of `emerge --resume` in the
> case of `emerge -e @world` compared to `emerge -uDN @world`. Sorry, if
> this was unclear.
>
> Heiko
>
>

I think I get what you are saying.  If for example you start a emerge -e
world, a emerge -uDN world or something and then stop it before it
finishes, running emerge --resume should pick up where you left off.  In
the past, I have done that after a reboot. I'm not sure if having some
things on tmpfs has a effect on that tho. 

That said, if you start one of those commands, emerge -e world for
example, and then do some other command besides --resume, then most
likely that will clear whatever emerge was doing before which means
--resume won't work because it has been reset/cleared with the second
command. 

As a workaround, I have been known to go to another terminal/konsole and
do a emerge --resume -a and let it get to the point where I need to hit
"y" and enter.  I let it sit there and go back to the original terminal
and emerge with whatever options I need for whatever package needs
attention.  Then when I'm done, I go to the other terminal/konsole and
tell emerge yes to the --resume command.  Once that command figures out
what it needs to do, it already has its list to work with.  However, I
can emerge something in another terminal to fix things and hopefully
carry on with the --resume. 

Sometimes doing that doesn't work but it could be worth a try.  It's
been a while since I've had the need to do that too.  Generally, if a
package fails, it will fail until something is fixed so that in can
complete the process.  As I've said before, emerge and how it does
things has come a long ways in recent years. 

I hope I understood what you meant with all this.  I disturbed quite a
few electrons and stuff with this.  lol

Dale

:-)  :-) 

Re: [gentoo-user] New profile 17: How urgent is the rebuild of world technically?

[Spackman, Chris]

On 2017/12/03 at 06:55am, Dale wrote:
 
> I think I get what you are saying.  If for example you start a
> emerge -e world, a emerge -uDN world or something and then stop it
> before it finishes, running emerge --resume should pick up where you
> left off.

Another helpful option, which I don't think has been mentioned yet, is
--skipfirst. With --resume, this is helpful when a relatively
unimportant package fails to compile. Emerge will skip the one that
failed (because it would be the first one in the resumed emerge) and
continue on. Later, I go back and see about getting the failed package
to work. I don't think that --skipfirst is a good idea if an important
package (one that will affect many other packages) fails. But, I am
not an expert on that stuff.

So, if:

emerge -e @world

fails (on a relatively unimportant package), you could use:

emerge --resume --skipfirst

to continue. I am actually almost 75% done with the system rebuild and
have had to do this so far with cdrdao and spideroak-bin (which
probably doesn't matter as it is a -bin package).

-- 
Chris Spackman

GNU Terry Pratchett

Re: [gentoo-user] New profile 17: How urgent is the rebuild of world technically?

[tuxic]

On 12/03 09:09, Spackman, Chris wrote:
> On 2017/12/03 at 06:55am, Dale wrote:
>  
> > I think I get what you are saying.  If for example you start a
> > emerge -e world, a emerge -uDN world or something and then stop it
> > before it finishes, running emerge --resume should pick up where you
> > left off.
> 
> Another helpful option, which I don't think has been mentioned yet, is
> --skipfirst. With --resume, this is helpful when a relatively
> unimportant package fails to compile. Emerge will skip the one that
> failed (because it would be the first one in the resumed emerge) and
> continue on. Later, I go back and see about getting the failed package
> to work. I don't think that --skipfirst is a good idea if an important
> package (one that will affect many other packages) fails. But, I am
> not an expert on that stuff.
> 
> So, if:
> 
> emerge -e @world
> 
> fails (on a relatively unimportant package), you could use:
> 
> emerge --resume --skipfirst
> 
> to continue. I am actually almost 75% done with the system rebuild and
> have had to do this so far with cdrdao and spideroak-bin (which
> probably doesn't matter as it is a -bin package).
> 
> -- 
> Chris Spackman
> 
> GNU Terry Pratchett
> 
> 

Hi,

what is about emerge -e @world --keep-going 
instead?

Cheers
Meino

Re: [gentoo-user] New profile 17: How urgent is the rebuild of world technically?

[Heiko Baums]

Am Sun, 3 Dec 2017 06:55:59 -0600
schrieb Dale <rdalek1967@gmail.com>:

> I hope I understood what you meant with all this.  I disturbed quite a
> few electrons and stuff with this.  lol

I think you understood what I meant even if I didn't think about doing
some other stuff with emerge in between another emerge. And I think
even Meino was more concerned about a power failure in between `emerge
-e @world`.

Nevertheless interesting to know that `emerge --resume` even works or
at least has once worked after another emerge.

Heiko

Re: [gentoo-user] New profile 17: How urgent is the rebuild of world technically?

[Heiko Baums]

Am Sun, 3 Dec 2017 09:09:37 -0500
schrieb "Spackman, Chris" <chris@osugisakae.com>:

> emerge --resume --skipfirst

`emerge --resume --skipfirst` is necessary if you don't use --keep-going
y, a package fails to build and you want to manually resume the actual
emerge. Not using --skipfirst wouldn't make much sense, because the
broken package will fail to build again anyway.

Maybe Dales suggestion would work here. In this case you shouldn't use
--skipfirst after fixing the reasons why the package failed to build.

If you run `emerge -e @world` e.g. and get a power failure then you
shouldn't use --skipfirst because then you want to build the package
which was currently built during the power failure again.

If you want to do have emerge doing a --resume --skipfirst
automatically then you should use --keep-going y in the original emerge
command like `emerge -e --keep-going y @world` or `emerge -uDN
--keep-going y @world`.

Heiko

Re: [gentoo-user] New profile 17: How urgent is the rebuild of world technically?

[Heiko Baums]

Am Sun, 3 Dec 2017 15:16:26 +0100
schrieb tuxic@posteo.de:

> what is about emerge -e @world --keep-going 
> instead?

That would do something like a --resume --skipfirst automatically with
the difference that it first recalculates the dependency tree in case
another package would depend on the package that failed to build.

But it's actually "--keep-going y".

Heiko

Re: [gentoo-user] New profile 17: How urgent is the rebuild of world technically?

[Peter Humphrey]

On Sunday, 3 December 2017 14:39:31 GMT Heiko Baums wrote:
> Am Sun, 3 Dec 2017 15:16:26 +0100
> 
> schrieb tuxic@posteo.de:
> > what is about emerge -e @world --keep-going
> > instead?
> 
> That would do something like a --resume --skipfirst automatically with
> the difference that it first recalculates the dependency tree in case
> another package would depend on the package that failed to build.
> 
> But it's actually "--keep-going y".

Depends where you specify it.

# alias emerj
alias emerj='emerge --jobs --load-average=36 --keep-going --nospinner'

I use that all the time; being a command on the command line, it overrides 
any environment values set in make.conf.

-- 
Regards,
Peter.

Re: [gentoo-user] New profile 17: How urgent is the rebuild of world technically?

[Neil Bothwick]

[-- Attachment #1: Type: text/plain, Size: 263 bytes --]

On Sun, 3 Dec 2017 15:39:31 +0100, Heiko Baums wrote:

> But it's actually "--keep-going y".

y is the default, so "--keep-going" and "--keep-going y" do the same
thing.


-- 
Neil Bothwick

Drive not ready: (R)etry (G)o to Impulse (C)all Engineering

[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 833 bytes --]

Re: [gentoo-user] New profile 17: How urgent is the rebuild of world technically?

[Dale]

Heiko Baums wrote:
> Am Sun, 3 Dec 2017 06:55:59 -0600
> schrieb Dale <rdalek1967@gmail.com>:
>
>> I hope I understood what you meant with all this.  I disturbed quite a
>> few electrons and stuff with this.  lol
> I think you understood what I meant even if I didn't think about doing
> some other stuff with emerge in between another emerge. And I think
> even Meino was more concerned about a power failure in between `emerge
> -e @world`.
>
> Nevertheless interesting to know that `emerge --resume` even works or
> at least has once worked after another emerge.
>
> Heiko
>
>


Just keep in mind, you have to start the resume in another
console/konsole first.  It's been a good while since I've had to do that
but it should work.  The biggest thing, starting the process so that the
resuming emerge already knows what to do before doing anything else in
another terminal.  That's what prevents it from clearing out what you
want to resume.

That said, if fixing something requires a USE flag change or some other
environmental change, all bets are off.  That will lead to other changes
that will not apply to the already loaded resume command. 

I hope that makes sense because it can be rather complicated if it
doesn't click as to what I'm describing. 

Based on all the threads, I'm sticking with the old profile until next
week or maybe two weeks.  Let some of this settle.  It seems things are
going pretty well but there does seem to be a few hiccups here and there. 

Dale

:-)  :-)

Re: [gentoo-user] New profile 17: How urgent is the rebuild of world technically?

[Heiko Baums]

Am Sun, 3 Dec 2017 19:08:25 -0600
schrieb Dale <rdalek1967@gmail.com>:

> I hope that makes sense because it can be rather complicated if it
> doesn't click as to what I'm describing. 

It does.

> Based on all the threads, I'm sticking with the old profile until next
> week or maybe two weeks.  Let some of this settle.  It seems things
> are going pretty well but there does seem to be a few hiccups here
> and there. 

I'm doing it right now on two PCs with --keep-going since almost one
day. At least one night more to go.

Some packages already failed to build but I don't know yet which. But
usually `emerge --keep-going` prints a list of the failed packages at
the end.

Heiko

Re: [gentoo-user] New profile 17: How urgent is the rebuild of world technically?

[Walter Dnes]

On Mon, Dec 04, 2017 at 02:18:30AM +0100, Heiko Baums wrote

> Some packages already failed to build but I don't know yet which. But
> usually `emerge --keep-going` prints a list of the failed packages at
> the end.

  If you've got it set up, try...

ll -rt /var/log/portage/elog/

-- 
Walter Dnes <waltdnes@waltdnes.org>
I don't run "desktop environments"; I run useful applications

Re: [gentoo-user] New profile 17: How urgent is the rebuild of world technically?

[Stefan G. Weichinger]

On 12/03/2017 03:30 AM, Michael Orlitzky wrote:

> However, you can delay switching to the new profile for a while.

For how long?

eselect news item tells me:

"Please migrate away from the 13.0 profiles within the six weeks after
GCC 6.4.0 has been stabilized on your architecture. The 13.0 profiles
will be deprecated then and removed in half a year."

As I see it gcc-6.4.0 is stable for amd64 already.

I have a number of servers out there running profile 13.0, having to
rebuild all their packages within the next 6 weeks has to be planned
somehow and needs time and thought IMO.

Re: [gentoo-user] New profile 17: How urgent is the rebuild of world technically?

[Michael Orlitzky]

On 12/04/2017 12:48 PM, Stefan G. Weichinger wrote:
> On 12/03/2017 03:30 AM, Michael Orlitzky wrote:
> 
>> However, you can delay switching to the new profile for a while.
> 
> For how long?
> 
> eselect news item tells me:
> 
> "Please migrate away from the 13.0 profiles within the six weeks after
> GCC 6.4.0 has been stabilized on your architecture. The 13.0 profiles
> will be deprecated then and removed in half a year."
> 

Once the profile is deprecated (not yet), you've got six months.

Keep in mind that a profile isn't actually all that complicated. It
consists mainly of a few small text files, and can likely be copied
locally just like you would with an ebuild.

You also aren't required to rebuild everything right now (although you
should, to get the PIE/SSP protection!). You can pause your "emerge -e
@world" at any point, and resume it during off-hours or a slow period.
So long as you don't need to build anything else in the meantime, you
can take as long as you want.

Re: [gentoo-user] New profile 17: How urgent is the rebuild of world technically?

[Stefan G. Weichinger]

Am 2017-12-04 um 21:21 schrieb Michael Orlitzky:

> Once the profile is deprecated (not yet), you've got six months.
> 
> Keep in mind that a profile isn't actually all that complicated. It
> consists mainly of a few small text files, and can likely be copied
> locally just like you would with an ebuild.
> 
> You also aren't required to rebuild everything right now (although you
> should, to get the PIE/SSP protection!). You can pause your "emerge -e
> @world" at any point, and resume it during off-hours or a slow period.
> So long as you don't need to build anything else in the meantime, you
> can take as long as you want.

This is helpful information, thanks, I can relax now ;-)

Re: [gentoo-user] New profile 17: How urgent is the rebuild of world technically?

[Frank Steinmetzger]

[-- Attachment #1: Type: text/plain, Size: 1144 bytes --]

On Sun, Dec 03, 2017 at 04:26:55AM +0100, tuxic@posteo.de wrote:

> If the compilation will fail at a certain point (and it will fail,
> since this is a complete new thing) -- would it be possible to resume
> even some tweaks, hacks and patches (even certain recompilations)
> would be needed in between?
> 
> Can I stop a running emerge @world and resume later?
> 
> How does a restarted emerge @world recognizes packages, which are
> already compiled according to the new standard?

I “circumvent” those questions by doing:
emerge -pveD world > worldlist
emerge -1O $(cat worldlist)

If the system for whatever reason fails and I need to interrupt the merge, I
simply remove the lines from worldlist that have already been built and then
repeat the last command. Plus I can exclude some packages that don’t need a
rebuild: -bins, -docs, virtuals, most perl and tex packages and so on. This
saves a bit of time on the slower laptop.

-- 
Gruß | Greetings | Qapla’
Please do not share anything from, with or about me on any social network.

This is a lousy party! I’ll leave as soon as I find my trousers.

[-- Attachment #2: Digital signature --]
[-- Type: application/pgp-signature, Size: 833 bytes --]

Re: [gentoo-user] New profile 17: How urgent is the rebuild of world technically?

[Neil Bothwick]

[-- Attachment #1: Type: text/plain, Size: 2425 bytes --]

On Thu, 7 Dec 2017 00:59:33 +0100, Frank Steinmetzger wrote:

> > How does a restarted emerge @world recognizes packages, which are
> > already compiled according to the new standard?  
> 
> I “circumvent” those questions by doing:
> emerge -pveD world > worldlist
> emerge -1O $(cat worldlist)
> 
> If the system for whatever reason fails and I need to interrupt the
> merge, I simply remove the lines from worldlist that have already been
> built and then repeat the last command. Plus I can exclude some
> packages that don’t need a rebuild: -bins, -docs, virtuals, most perl
> and tex packages and so on. This saves a bit of time on the slower
> laptop.

I wrote a script to handle this some years ago, and it has come in handy
this week. It emerges all packages that have not been done since a given
time.

In this case, I run

mergeolderthan -r glibc

since glibc was emerged right before the world emerge

#!/bin/bash

EMERGE_ARGS="--oneshot --keep-going"

usage() {
	echo -e "\nUsage: $(basename $0) [-f file] [-r category/package[-version] [-h]"
	echo "    -f re-emerge all packages older than this file"
	echo "    -r re-emerge all packages older than this package"
	echo "    -h Show this text"
	echo -e "\n    All other options are passed to the emerge command"
	echo -e "$*"
	exit
	}

while getopts f:r:pvlh ARG; do
	case "${ARG}" in
		f) REFFILE=${OPTARG} ;;
		r) REFFILE=$(ls -1 /var/db/pkg/${OPTARG}*/environment.bz2 | head -n 1) ;;
		p) EMERGE_ARGS="${EMERGE_ARGS} --pretend" ;;
		v) EMERGE_ARGS="${EMERGE_ARGS} --verbose" ;;
		l) LIST="y" ;;
		h) usage ;;
		esac
	done
shift $(expr ${OPTIND} - 1)

[[ "${REFFILE}" ]] || usage "\nYou must specify a reference with -f or -r\n"
[[ -f ${REFFILE} ]] || usage "\n${REFFILE} not found\n"

PKGLIST=$(mktemp -t mergeolderthan.XXXXXXXX)

emerge -ep --exclude gentoo-sources @world | grep -v sys-kernel/gentoo-sources | awk -F] '/^\[ebuild/ {print $2}' | awk '{print $1}' | while read PKG; do
	if [[ /var/db/pkg/${PKG}/environment.bz2 -ot ${REFFILE} ]]; then
		echo "=${PKG}" >>$PKGLIST
		fi
	done

if [[ "${LIST}" ]]; then
	cat ${PKGLIST} && rm -f ${PKGLIST}
else
	cat ${PKGLIST} | xargs --no-run-if-empty emerge ${EMERGE_ARGS} && rm -f ${PKGLIST}
	fi



-- 
Neil Bothwick

Hickory Dickory Dock, The mice ran up the clock, The clock struck one, The
others escaped with minor injuries.

[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 833 bytes --]