[gentoo-user] RAM checks for chromium

[gentoo-user] RAM checks for chromium

[Mick]

[-- Attachment #1: Type: text/plain, Size: 747 bytes --]

Two Intel systems with 4G RAM failed to build chromium, even after setting 
MAKEOPTS="-j2". The ebuild is checking for a minimum of 3G RAM:

>>> Running pre-merge checks for www-client/chromium-70.0.3538.110
 * Checking for at least 3 GiB RAM ...                                [ ok ]
 * Checking for at least 5 GiB disk space at "/var/tmp/portage/www-client/
chromium-70.0.3538.110/temp" ...                                      [ ok ]

Given I've spent more than two days compiling to get nowhere with this, I'm 
thinking:

a) Chromium probably needs more than 3G now.
b) Either the ebuild, or portage, ought to check available RAM and dynamically 
adjust the number of jobs accordingly - or have I watched too many AI movies?

-- 
Regards,
Mick

[-- Attachment #2: This is a digitally signed message part. --]
[-- Type: application/pgp-signature, Size: 833 bytes --]

Re: [gentoo-user] RAM checks for chromium

[Alexander Kapshuk]

On Tue, Dec 4, 2018 at 9:35 AM Mick <michaelkintzios@gmail.com> wrote:
>
> Two Intel systems with 4G RAM failed to build chromium, even after setting
> MAKEOPTS="-j2". The ebuild is checking for a minimum of 3G RAM:
>
> >>> Running pre-merge checks for www-client/chromium-70.0.3538.110
>  * Checking for at least 3 GiB RAM ...                                [ ok ]
>  * Checking for at least 5 GiB disk space at "/var/tmp/portage/www-client/
> chromium-70.0.3538.110/temp" ...                                      [ ok ]
>
> Given I've spent more than two days compiling to get nowhere with this, I'm
> thinking:
>
> a) Chromium probably needs more than 3G now.
> b) Either the ebuild, or portage, ought to check available RAM and dynamically
> adjust the number of jobs accordingly - or have I watched too many AI movies?
>
> --
> Regards,
> Mick

You're right. Chromium does require more than 3G of RAM to build.
Here are the current system requirements for building Chromium on Linux:

System requirements
A 64-bit Intel machine with at least 8GB of RAM. More than 16GB is
highly recommended.
At least 100GB of free disk space.
You must have Git and Python v2 installed already.

See the link below for details.
https://chromium.googlesource.com/chromium/src/+/HEAD/docs/linux_build_instructions.md#system-requirements

Re: [gentoo-user] RAM checks for chromium

[Mick]

[-- Attachment #1: Type: text/plain, Size: 1931 bytes --]

On Tuesday, 4 December 2018 08:06:22 GMT Alexander Kapshuk wrote:
> On Tue, Dec 4, 2018 at 9:35 AM Mick <michaelkintzios@gmail.com> wrote:
> > Two Intel systems with 4G RAM failed to build chromium, even after setting
> > 
> > MAKEOPTS="-j2". The ebuild is checking for a minimum of 3G RAM:
> > >>> Running pre-merge checks for www-client/chromium-70.0.3538.110
> >  
> >  * Checking for at least 3 GiB RAM ...                                [ ok
> >  ]
> >  * Checking for at least 5 GiB disk space at "/var/tmp/portage/www-client/
> > 
> > chromium-70.0.3538.110/temp" ...                                      [ ok
> > ]
> > 
> > Given I've spent more than two days compiling to get nowhere with this,
> > I'm
> > thinking:
> > 
> > a) Chromium probably needs more than 3G now.
> > b) Either the ebuild, or portage, ought to check available RAM and
> > dynamically adjust the number of jobs accordingly - or have I watched too
> > many AI movies?
> > 
> > --
> > Regards,
> > Mick
> 
> You're right. Chromium does require more than 3G of RAM to build.
> Here are the current system requirements for building Chromium on Linux:
> 
> System requirements
> A 64-bit Intel machine with at least 8GB of RAM. More than 16GB is
> highly recommended.

OK it figures, an AMD system with 16G RAM and /var/portage/ on a tmpfs had no 
problem.


> At least 100GB of free disk space.

O_O  What the ... ?


> You must have Git and Python v2 installed already.
> 
> See the link below for details.
> https://chromium.googlesource.com/chromium/src/+/HEAD/docs/linux_build_instr
> uctions.md#system-requirements

Thanks for this.  It may be I'll need to build chromium as a binary on the 
faster PC from now on and copy it over to the older clients, but I can't 
recall what command spews out the detailed CFLAGS for the client which I will 
need to run on the faster host's CLI to emerge the binary.  Grateful for any 
hints.

-- 
Regards,
Mick

[-- Attachment #2: This is a digitally signed message part. --]
[-- Type: application/pgp-signature, Size: 833 bytes --]

Re: [gentoo-user] RAM checks for chromium

[Raffaele Belardi]

> Thanks for this.  It may be I'll need to build chromium as a binary on the
> faster PC from now on and copy it over to the older clients, but I can't
> recall what command spews out the detailed CFLAGS for the client which I will
> need to run on the faster host's CLI to emerge the binary.  Grateful for any
> hints.

app-portage/cpuid2cpuflags

Mick wrote:
> On Tuesday, 4 December 2018 08:06:22 GMT Alexander Kapshuk wrote:
>> On Tue, Dec 4, 2018 at 9:35 AM Mick <michaelkintzios@gmail.com> wrote:
>>> Two Intel systems with 4G RAM failed to build chromium, even after setting
>>>
>>> MAKEOPTS="-j2". The ebuild is checking for a minimum of 3G RAM:
>>>>>> Running pre-merge checks for www-client/chromium-70.0.3538.110
>>>  
>>>  * Checking for at least 3 GiB RAM ...                                [ ok
>>>  ]
>>>  * Checking for at least 5 GiB disk space at "/var/tmp/portage/www-client/
>>>
>>> chromium-70.0.3538.110/temp" ...                                      [ ok
>>> ]
>>>
>>> Given I've spent more than two days compiling to get nowhere with this,
>>> I'm
>>> thinking:
>>>
>>> a) Chromium probably needs more than 3G now.
>>> b) Either the ebuild, or portage, ought to check available RAM and
>>> dynamically adjust the number of jobs accordingly - or have I watched too
>>> many AI movies?
>>>
>>> --
>>> Regards,
>>> Mick
>>
>> You're right. Chromium does require more than 3G of RAM to build.
>> Here are the current system requirements for building Chromium on Linux:
>>
>> System requirements
>> A 64-bit Intel machine with at least 8GB of RAM. More than 16GB is
>> highly recommended.
> 
> OK it figures, an AMD system with 16G RAM and /var/portage/ on a tmpfs had no 
> problem.
> 
> 
>> At least 100GB of free disk space.
> 
> O_O  What the ... ?
> 
> 
>> You must have Git and Python v2 installed already.
>>
>> See the link below for details.
>> https://chromium.googlesource.com/chromium/src/+/HEAD/docs/linux_build_instr
>> uctions.md#system-requirements
> 
> Thanks for this.  It may be I'll need to build chromium as a binary on the 
> faster PC from now on and copy it over to the older clients, but I can't 
> recall what command spews out the detailed CFLAGS for the client which I will 
> need to run on the faster host's CLI to emerge the binary.  Grateful for any 
> hints.
> 

Re: [gentoo-user] RAM checks for chromium

[Alexander Kapshuk]

On Tue, Dec 4, 2018 at 10:47 AM Mick <michaelkintzios@gmail.com> wrote:
>
> On Tuesday, 4 December 2018 08:06:22 GMT Alexander Kapshuk wrote:
> > On Tue, Dec 4, 2018 at 9:35 AM Mick <michaelkintzios@gmail.com> wrote:
> > > Two Intel systems with 4G RAM failed to build chromium, even after setting
> > >
> > > MAKEOPTS="-j2". The ebuild is checking for a minimum of 3G RAM:
> > > >>> Running pre-merge checks for www-client/chromium-70.0.3538.110
> > >
> > >  * Checking for at least 3 GiB RAM ...                                [ ok
> > >  ]
> > >  * Checking for at least 5 GiB disk space at "/var/tmp/portage/www-client/
> > >
> > > chromium-70.0.3538.110/temp" ...                                      [ ok
> > > ]
> > >
> > > Given I've spent more than two days compiling to get nowhere with this,
> > > I'm
> > > thinking:
> > >
> > > a) Chromium probably needs more than 3G now.
> > > b) Either the ebuild, or portage, ought to check available RAM and
> > > dynamically adjust the number of jobs accordingly - or have I watched too
> > > many AI movies?
> > >
> > > --
> > > Regards,
> > > Mick
> >
> > You're right. Chromium does require more than 3G of RAM to build.
> > Here are the current system requirements for building Chromium on Linux:
> >
> > System requirements
> > A 64-bit Intel machine with at least 8GB of RAM. More than 16GB is
> > highly recommended.
>
> OK it figures, an AMD system with 16G RAM and /var/portage/ on a tmpfs had no
> problem.
>
>
> > At least 100GB of free disk space.
>
> O_O  What the ... ?
>
>
> > You must have Git and Python v2 installed already.
> >
> > See the link below for details.
> > https://chromium.googlesource.com/chromium/src/+/HEAD/docs/linux_build_instr
> > uctions.md#system-requirements
>
> Thanks for this.  It may be I'll need to build chromium as a binary on the
> faster PC from now on and copy it over to the older clients, but I can't
> recall what command spews out the detailed CFLAGS for the client which I will
> need to run on the faster host's CLI to emerge the binary.  Grateful for any
> hints.
>
> --
> Regards,
> Mick

Perhaps these two gcc commands are what you're after:
gcc -c -Q -march=native --help=target
gcc -### -march=native /usr/include/stdlib.h

See this link for details, https://wiki.gentoo.org/wiki/GCC_optimization

Re: [gentoo-user] RAM checks for chromium

[Jack]

On 2018.12.04 02:35, Mick wrote:
> Two Intel systems with 4G RAM failed to build chromium, even after  
> setting
> MAKEOPTS="-j2". The ebuild is checking for a minimum of 3G RAM:
> 
> >>> Running pre-merge checks for www-client/chromium-70.0.3538.110
>  * Checking for at least 3 GiB RAM ...                                 
> [ ok ]
>  * Checking for at least 5 GiB disk space at  
> "/var/tmp/portage/www-client/
> chromium-70.0.3538.110/temp" ...                                       
> [ ok ]
> 
> Given I've spent more than two days compiling to get nowhere with  
> this, I'm
> thinking:
One thing I've done in the past if something failed after a long time  
compiling is to cd to the top build dir (under the Portage tmp dir) and  
just continue the compile (either make or ninja, or whatever that  
package uses) when/if that finishes, you can use ebuild to finish the  
install and qmerge steps.  That avoids needing to start the compile  
from the beginning.
> 
> a) Chromium probably needs more than 3G now.
> b) Either the ebuild, or portage, ought to check available RAM and  
> dynamically
> adjust the number of jobs accordingly - or have I watched too many AI  
> movies?
> 
> --
> Regards,
> Mick

Jack

Re: [gentoo-user] RAM checks for chromium

[Neil Bothwick]

[-- Attachment #1: Type: text/plain, Size: 841 bytes --]

On Tue, 04 Dec 2018 13:42:16 -0500, Jack wrote:

> > Given I've spent more than two days compiling to get nowhere with  
> > this, I'm
> > thinking:  
> One thing I've done in the past if something failed after a long time  
> compiling is to cd to the top build dir (under the Portage tmp dir)
> and just continue the compile (either make or ninja, or whatever that  
> package uses) when/if that finishes, you can use ebuild to finish the  
> install and qmerge steps.  That avoids needing to start the compile  
> from the beginning.

You can use ebuild for that too, with the compile option. I've have the
chromium build fail for apparently random reasons on my laptop from time
to time and ebuild ... compile finishes the process.


-- 
Neil Bothwick

-Come, come, why they couldn't hit an elephant from this dist-

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

Re: [gentoo-user] RAM checks for chromium

[Jack]

On 2018.12.04 14:13, Neil Bothwick wrote:
> On Tue, 04 Dec 2018 13:42:16 -0500, Jack wrote:
> 
>>> Given I've spent more than two days compiling to get nowhere with  
>>> this, I'm thinking:
>> One thing I've done in the past if something failed after a long  
>> time compiling is to cd to the top build dir (under the Portage tmp  
>> dir) and just continue the compile (either make or ninja, or  
>> whatever that package uses) when/if that finishes, you can use  
>> ebuild to finish the install and qmerge steps.  That avoids needing  
>> to start the compile from the beginning.
> You can use ebuild for that too, with the compile option. I've have  
> the chromium build fail for apparently random reasons on my laptop  
> from time to time and ebuild ... compile finishes the process.

Unless I'm mistaken, "ebuild /path/to/ebuild compile" does avoid  
redoing the unpack, prepare, and configure steps, but it  starts the  
compile from scratch.  Manually doing "make" (or whatever) in the  
appropriate directory avoids repeating those parts of the compile that  
were successful.  If the compile takes two days, that's a significant  
savings in time.

Jack

Re: [gentoo-user] RAM checks for chromium

[Mick]

[-- Attachment #1: Type: text/plain, Size: 1818 bytes --]

On Tuesday, 4 December 2018 19:23:27 GMT Jack wrote:
> On 2018.12.04 14:13, Neil Bothwick wrote:
> > On Tue, 04 Dec 2018 13:42:16 -0500, Jack wrote:
> >>> Given I've spent more than two days compiling to get nowhere with
> >> 
> >>> this, I'm thinking:
> >> One thing I've done in the past if something failed after a long
> >> time compiling is to cd to the top build dir (under the Portage tmp
> >> dir) and just continue the compile (either make or ninja, or
> >> whatever that package uses) when/if that finishes, you can use
> >> ebuild to finish the install and qmerge steps.  That avoids needing
> >> to start the compile from the beginning.
> > 
> > You can use ebuild for that too, with the compile option. I've have
> > the chromium build fail for apparently random reasons on my laptop
> > from time to time and ebuild ... compile finishes the process.
> 
> Unless I'm mistaken, "ebuild /path/to/ebuild compile" does avoid
> redoing the unpack, prepare, and configure steps, but it  starts the
> compile from scratch.  Manually doing "make" (or whatever) in the
> appropriate directory avoids repeating those parts of the compile that
> were successful.  If the compile takes two days, that's a significant
> savings in time.
> 
> Jack

Phew!  The chromium emerge completed with -j1, although it took 4 hours longer 
than last time on one PC and 6.5 hours longer on another.  I hope one day 
ebuilds/compilers will be clever enough to deduce a suitable job number a 
system will run without thrashing swap aimlessly, on a per package basis.  
Some versions of chromium compile fine with -j3, others seem to be so hungry 
on RAM they would need -j2.  I think this must be the first time with chromium 
I had to drop the job number down to 1.

Thank you again for all the pointers and ideas.
-- 
Regards,
Mick

[-- Attachment #2: This is a digitally signed message part. --]
[-- Type: application/pgp-signature, Size: 833 bytes --]

Re: [gentoo-user] RAM checks for chromium

[Adam Carter]

[-- Attachment #1: Type: text/plain, Size: 519 bytes --]

On Wed, Dec 5, 2018 at 7:41 AM Mick <michaelkintzios@gmail.com> wrote:

> On Tuesday, 4 December 2018 19:23:27 GMT Jack wrote:
> Phew!  The chromium emerge completed with -j1, although it took 4 hours
> longer
> than last time on one PC and 6.5 hours longer on another.
>

For those systems it might be worth trying the binary google-chrome
instead. Much smaller download and;

$ genlop -t google-chrome | tail -n3
     Tue Nov 20 20:20:10 2018 >>> www-client/google-chrome-70.0.3538.110
       merge time: 35 seconds.

[-- Attachment #2: Type: text/html, Size: 894 bytes --]

Re: [gentoo-user] RAM checks for chromium

[Jack]

On 2018.12.04 20:36, Adam Carter wrote:
> On Wed, Dec 5, 2018 at 7:41 AM Mick <michaelkintzios@gmail.com> wrote:
> 
> > On Tuesday, 4 December 2018 19:23:27 GMT Jack wrote:
> > Phew!  The chromium emerge completed with -j1, although it took 4  
> hours
> > longer
> > than last time on one PC and 6.5 hours longer on another.
> >
> 
> For those systems it might be worth trying the binary google-chrome
> instead. Much smaller download and;
> 
> $ genlop -t google-chrome | tail -n3
>      Tue Nov 20 20:20:10 2018 >>>  
> www-client/google-chrome-70.0.3538.110
>        merge time: 35 seconds.
> 
But only if you don't care about the differences between Chrome and  
Chromium.  They are close, but not (unless I'm terribly mistaken)  
exactly the same.  The latter is completely FOSS, but the former  
contains some Google specific additions.  While I do have both  
installed, I don't particularly trust Google enough to use Chrome  
unless nothing else works.  It's pretty rare I need to use it.

Jack

Re: [gentoo-user] RAM checks for chromium

[Mick]

[-- Attachment #1: Type: text/plain, Size: 1517 bytes --]

On Wednesday, 5 December 2018 03:26:50 GMT Jack wrote:
> On 2018.12.04 20:36, Adam Carter wrote:
> > On Wed, Dec 5, 2018 at 7:41 AM Mick <michaelkintzios@gmail.com> wrote:
> > > On Tuesday, 4 December 2018 19:23:27 GMT Jack wrote:
> > > Phew!  The chromium emerge completed with -j1, although it took 4
> > 
> > hours
> > 
> > > longer
> > > than last time on one PC and 6.5 hours longer on another.
> > 
> > For those systems it might be worth trying the binary google-chrome
> > instead. Much smaller download and;
> > 
> > $ genlop -t google-chrome | tail -n3
> > 
> >      Tue Nov 20 20:20:10 2018 >>>
> > 
> > www-client/google-chrome-70.0.3538.110
> > 
> >        merge time: 35 seconds.
> 
> But only if you don't care about the differences between Chrome and
> Chromium.  They are close, but not (unless I'm terribly mistaken)
> exactly the same.  The latter is completely FOSS, but the former
> contains some Google specific additions.  While I do have both
> installed, I don't particularly trust Google enough to use Chrome
> unless nothing else works.  It's pretty rare I need to use it.
> 
> Jack

I chose Chromium because I understood Google-Chrome to have some settings/
code, which is meant to link Google services with a user's footprint even 
after you have logged out of all Google services.

What I found with Chromium is that it varies considerably on the amount of RAM 
consumed between versions, so it is a matter of guessing if 4G of RAM would be 
adequate to support > -j1.

-- 
Regards,
Mick

[-- Attachment #2: This is a digitally signed message part. --]
[-- Type: application/pgp-signature, Size: 833 bytes --]

Re: [gentoo-user] RAM checks for chromium

[Peter Humphrey]

On Wednesday, 5 December 2018 09:16:10 GMT Mick wrote:

> I chose Chromium because I understood Google-Chrome to have some settings/
> code, which is meant to link Google services with a user's footprint even
> after you have logged out of all Google services.
> 
> What I found with Chromium is that it varies considerably on the amount of
> RAM consumed between versions, so it is a matter of guessing if 4G of RAM
> would be adequate to support > -j1.

I seem to remember chrome pulling a few dependencies that chromium doesn't.

The main reason I've ditched chrome and chromium altogether is that they 
insist on redirecting me to their mobile site - and this is a 27-inch screen! 
But I'm also uncomfortable with the privacy concerns such as you mention.

-- 
Regards,
Peter.

Re: [gentoo-user] RAM checks for chromium

[Mick]

[-- Attachment #1: Type: text/plain, Size: 452 bytes --]

On Wednesday, 5 December 2018 10:12:10 GMT Peter Humphrey wrote:
> The main reason I've ditched chrome and chromium altogether is that they
> insist on redirecting me to their mobile site - and this is a 27-inch
> screen! But I'm also uncomfortable with the privacy concerns such as you
> mention.
> 
> --
> Regards,
> Peter.

I haven't noticed something like this here.  Is there a particular site that 
causes this you could share?

-- 
Regards,
Mick

[-- Attachment #2: This is a digitally signed message part. --]
[-- Type: application/pgp-signature, Size: 833 bytes --]

Re: [gentoo-user] RAM checks for chromium

[Peter Humphrey]

On Wednesday, 5 December 2018 11:11:06 GMT Mick wrote:
> On Wednesday, 5 December 2018 10:12:10 GMT Peter Humphrey wrote:
> > The main reason I've ditched chrome and chromium altogether is that they
> > insist on redirecting me to their mobile site - and this is a 27-inch
> > screen! But I'm also uncomfortable with the privacy concerns such as you
> > mention.
> 
> I haven't noticed something like this here.  Is there a particular site that
> causes this you could share?

I'd like to check, but today I can't compile chromium; it throws a lot of 
errors about an end of file coming before the end of data. I suppose it's 
running out of space, though there seems to be plenty. I'll look into it.

As far as I can remember though, it was the google.com home page. I didn't get 
any further than that before complaining to google, waiting for the inevitably 
absent reply, then giving up and ditching it.

-- 
Regards,
Peter.

Re: [gentoo-user] RAM checks for chromium

[Peter Humphrey]

On Thursday, 6 December 2018 10:35:19 GMT Peter Humphrey wrote:
> On Wednesday, 5 December 2018 11:11:06 GMT Mick wrote:
> > On Wednesday, 5 December 2018 10:12:10 GMT Peter Humphrey wrote:
> > > The main reason I've ditched chrome and chromium altogether is that they
> > > insist on redirecting me to their mobile site - and this is a 27-inch
> > > screen! But I'm also uncomfortable with the privacy concerns such as you
> > > mention.
> > 
> > I haven't noticed something like this here.  Is there a particular site
> > that causes this you could share?
> 
> I'd like to check, but today I can't compile chromium; it throws a lot of
> errors about an end of file coming before the end of data. I suppose it's
> running out of space, though there seems to be plenty. I'll look into it.
> 
> As far as I can remember though, it was the google.com home page. I didn't
> get any further than that before complaining to google, waiting for the
> inevitably absent reply, then giving up and ditching it.

Well, I did manage to compile chromium in 165 minutes, but I had to switch off 
jumbo-build to do it. I'd have thought that 32GB RAM and 38GB swap would be 
enough for anything reasonable, but perhaps chromium isn't reasonable.

And today, of course, there's an upgrade. That's another reason I ditched it. 
Is there a way to force chromium to be not ~amd64 on a ~amd64 system?

-- 
Regards,
Peter.

Re: [gentoo-user] RAM checks for chromium

[Andreas Fink]

On Fri, 07 Dec 2018 10:14:45 +0000
Peter Humphrey <peter@prh.myzen.co.uk> wrote:

> And today, of course, there's an upgrade. That's another reason I
> ditched it. Is there a way to force chromium to be not ~amd64 on a
> ~amd64 system?

Yes, I do that with this entry in /etc/portage/package.keywords:
www-client/chromium -~amd64

Cheers
Andreas

Re: [gentoo-user] RAM checks for chromium

[Peter Humphrey]

On Friday, 7 December 2018 10:17:17 GMT Andreas Fink wrote:
> On Fri, 07 Dec 2018 10:14:45 +0000
> 
> Peter Humphrey <peter@prh.myzen.co.uk> wrote:
> > And today, of course, there's an upgrade. That's another reason I
> > ditched it. Is there a way to force chromium to be not ~amd64 on a
> > ~amd64 system?
> 
> Yes, I do that with this entry in /etc/portage/package.keywords:
> www-client/chromium -~amd64

Ah, yes, of course. Ta muchly.

-- 
Regards,
Peter.

Re: [gentoo-user] RAM checks for chromium

[Rich Freeman]

On Fri, Dec 7, 2018 at 5:48 AM Peter Humphrey <peter@prh.myzen.co.uk> wrote:
>
> On Friday, 7 December 2018 10:17:17 GMT Andreas Fink wrote:
> > On Fri, 07 Dec 2018 10:14:45 +0000
> >
> > Peter Humphrey <peter@prh.myzen.co.uk> wrote:
> > > And today, of course, there's an upgrade. That's another reason I
> > > ditched it. Is there a way to force chromium to be not ~amd64 on a
> > > ~amd64 system?
> >
> > Yes, I do that with this entry in /etc/portage/package.keywords:
> > www-client/chromium -~amd64
>
> Ah, yes, of course. Ta muchly.
>

You'll be doing a build shortly in any case:
https://bugs.gentoo.org/672606

So, I'd hold off on any upgrades/downgrades/etc for a day or so until
this hits if CPU time matters to you.

-- 
Rich

Re: [gentoo-user] RAM checks for chromium

[Peter Humphrey]

On Friday, 7 December 2018 15:01:09 GMT Rich Freeman wrote:
> On Fri, Dec 7, 2018 at 5:48 AM Peter Humphrey <peter@prh.myzen.co.uk> wrote:
> > On Friday, 7 December 2018 10:17:17 GMT Andreas Fink wrote:
> > > On Fri, 07 Dec 2018 10:14:45 +0000
> > > 
> > > Peter Humphrey <peter@prh.myzen.co.uk> wrote:
> > > > And today, of course, there's an upgrade. That's another reason I
> > > > ditched it. Is there a way to force chromium to be not ~amd64 on a
> > > > ~amd64 system?
> > > 
> > > Yes, I do that with this entry in /etc/portage/package.keywords:
> > > www-client/chromium -~amd64
> > 
> > Ah, yes, of course. Ta muchly.
> 
> You'll be doing a build shortly in any case:
> https://bugs.gentoo.org/672606
> 
> So, I'd hold off on any upgrades/downgrades/etc for a day or so until
> this hits if CPU time matters to you.

Too late! was the cry...

Never mind; that's interesting anyway. 43 security fixes in one release, eh? 
Perhaps we should all be wary of the beast until the other umpteen have been 
fixed as well.

-- 
Regards,
Peter.

Re: [gentoo-user] RAM checks for chromium

[Neil Bothwick]

[-- Attachment #1: Type: text/plain, Size: 1448 bytes --]

On Tue, 04 Dec 2018 14:23:27 -0500, Jack wrote:

> On 2018.12.04 14:13, Neil Bothwick wrote:

> >> One thing I've done in the past if something failed after a long  
> >> time compiling is to cd to the top build dir (under the Portage tmp  
> >> dir) and just continue the compile (either make or ninja, or  
> >> whatever that package uses) when/if that finishes, you can use  
> >> ebuild to finish the install and qmerge steps.  That avoids needing  
> >> to start the compile from the beginning.  
> > You can use ebuild for that too, with the compile option. I've have  
> > the chromium build fail for apparently random reasons on my laptop  
> > from time to time and ebuild ... compile finishes the process.  
> 
> Unless I'm mistaken, "ebuild /path/to/ebuild compile" does avoid  
> redoing the unpack, prepare, and configure steps, but it  starts the  
> compile from scratch.  Manually doing "make" (or whatever) in the  
> appropriate directory avoids repeating those parts of the compile that  
> were successful.  If the compile takes two days, that's a significant  
> savings in time.

It starts the compile by running make or whatever is appropriate for the
build, so it doesn't need to build anything already built any more than a
bare make does. But using ebuild compile means you get the same
environment as when you started the compile.


-- 
Neil Bothwick

Why is the word abbreviation so long?

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

Re: [gentoo-user] RAM checks for chromium

[Mick]

[-- Attachment #1: Type: text/plain, Size: 1662 bytes --]

On Tuesday, 4 December 2018 20:57:11 GMT Neil Bothwick wrote:
> On Tue, 04 Dec 2018 14:23:27 -0500, Jack wrote:
> > On 2018.12.04 14:13, Neil Bothwick wrote:
> > >> One thing I've done in the past if something failed after a long
> > >> time compiling is to cd to the top build dir (under the Portage tmp
> > >> dir) and just continue the compile (either make or ninja, or
> > >> whatever that package uses) when/if that finishes, you can use
> > >> ebuild to finish the install and qmerge steps.  That avoids needing
> > >> to start the compile from the beginning.
> > > 
> > > You can use ebuild for that too, with the compile option. I've have
> > > the chromium build fail for apparently random reasons on my laptop
> > > from time to time and ebuild ... compile finishes the process.
> > 
> > Unless I'm mistaken, "ebuild /path/to/ebuild compile" does avoid
> > redoing the unpack, prepare, and configure steps, but it  starts the
> > compile from scratch.  Manually doing "make" (or whatever) in the
> > appropriate directory avoids repeating those parts of the compile that
> > were successful.  If the compile takes two days, that's a significant
> > savings in time.
> 
> It starts the compile by running make or whatever is appropriate for the
> build, so it doesn't need to build anything already built any more than a
> bare make does. But using ebuild compile means you get the same
> environment as when you started the compile.

Which will be counterproductive if the reason the compile failed is because 
RAM was exhausted and you need to reduce the job number.  Could I define 
MAKEOPTS on the CLI when running ebuild by hand?

-- 
Regards,
Mick

[-- Attachment #2: This is a digitally signed message part. --]
[-- Type: application/pgp-signature, Size: 833 bytes --]

Re: [gentoo-user] RAM checks for chromium

[Neil Bothwick]

[-- Attachment #1: Type: text/plain, Size: 648 bytes --]

On Tue, 04 Dec 2018 21:10:05 +0000, Mick wrote:

> > It starts the compile by running make or whatever is appropriate for
> > the build, so it doesn't need to build anything already built any
> > more than a bare make does. But using ebuild compile means you get
> > the same environment as when you started the compile.  
> 
> Which will be counterproductive if the reason the compile failed is
> because RAM was exhausted and you need to reduce the job number.  Could
> I define MAKEOPTS on the CLI when running ebuild by hand?

Yes.


-- 
Neil Bothwick

Suborbital Ballistic-Propulsion Engineer
Not Exactly A Rocket Scientist

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