We'll stop testing i586 for TW

classic Classic list List threaded Threaded
11 messages Options
Reply | Threaded
Open this post in threaded view
|

We'll stop testing i586 for TW

Stephan Kulow-3
Hi,

As we are under heavy load on openqa.opensuse.org with 2 releases and
its stagings, we'll stop testing the i586 ISOs for TW. I won't change
anything about them being built for now, but the next step would be
to publish these ISOs somewhere outside of /tumbleweed/iso.

I'm not so sure about the timeline we should follow with this, but as
you might know by now, my interest in historical platforms isn't high.

Gretings, Stephan

--
Ma muaß weiterkämpfen, kämpfen bis zum Umfalln, a wenn die
ganze Welt an Arsch offen hat, oder grad deswegn.
--
To unsubscribe, e-mail: [hidden email]
To contact the owner, e-mail: [hidden email]

Reply | Threaded
Open this post in threaded view
|

Re: We'll stop testing i586 for TW

Ondřej Súkup
+1

On 16 October 2015 at 11:17, Stephan Kulow <[hidden email]> wrote:

> Hi,
>
> As we are under heavy load on openqa.opensuse.org with 2 releases and
> its stagings, we'll stop testing the i586 ISOs for TW. I won't change
> anything about them being built for now, but the next step would be
> to publish these ISOs somewhere outside of /tumbleweed/iso.
>
> I'm not so sure about the timeline we should follow with this, but as
> you might know by now, my interest in historical platforms isn't high.
>
> Gretings, Stephan
>
> --
> Ma muaß weiterkämpfen, kämpfen bis zum Umfalln, a wenn die
> ganze Welt an Arsch offen hat, oder grad deswegn.
> --
> To unsubscribe, e-mail: [hidden email]
> To contact the owner, e-mail: [hidden email]
>
--
To unsubscribe, e-mail: [hidden email]
To contact the owner, e-mail: [hidden email]

Reply | Threaded
Open this post in threaded view
|

Re: We'll stop testing i586 for TW

Olaf Hering-2
In reply to this post by Stephan Kulow-3
On Fri, Oct 16, Stephan Kulow wrote:

> I'm not so sure about the timeline we should follow with this, but as
> you might know by now, my interest in historical platforms isn't high.

The next step should be to stop building anything without a
baselibs.conf in the x86 repos.

Olaf
--
To unsubscribe, e-mail: [hidden email]
To contact the owner, e-mail: [hidden email]

Reply | Threaded
Open this post in threaded view
|

Re: We'll stop testing i586 for TW

Stephan Kulow-3
Am 16.10.2015 um 11:35 schrieb Olaf Hering:
> On Fri, Oct 16, Stephan Kulow wrote:
>
>> I'm not so sure about the timeline we should follow with this, but as
>> you might know by now, my interest in historical platforms isn't high.
>
> The next step should be to stop building anything without a
> baselibs.conf in the x86 repos.
>
Not that easy dude - e.g. you need automake for some of the above. But
if you have a bot ready to handle such cases, I would love to enable it
for Leap actually. I hate to wait for i586 libreoffice ;(

Greetings, Stephan

--
Ma muaß weiterkämpfen, kämpfen bis zum Umfalln, a wenn die
ganze Welt an Arsch offen hat, oder grad deswegn.
--
To unsubscribe, e-mail: [hidden email]
To contact the owner, e-mail: [hidden email]

Reply | Threaded
Open this post in threaded view
|

Re: We'll stop testing i586 for TW

Olaf Hering-2
Am 16.10.2015 um 11:49 schrieb Stephan Kulow:
> Not that easy dude - e.g. you need automake for some of the above. But
> if you have a bot ready to handle such cases, I would love to enable it
> for Leap actually. I hate to wait for i586 libreoffice ;(

I did not mean to do it now, just when the time has come.

Not sure how to automate it. Either x86 gets globally disabled, and the
base chroot + each pkg with baselibs gets enabled. Or x86 remains
globablly enabled, and each pkg without baselibs - base chroot gets
disabled.

Shouldnt it be like "obs ls -e $prj $pkg baselibs.conf && enable()"?

Olaf
--
To unsubscribe, e-mail: [hidden email]
To contact the owner, e-mail: [hidden email]

Reply | Threaded
Open this post in threaded view
|

Re: We'll stop testing i586 for TW

Jan Engelhardt-4
On Friday 2015-10-16 11:56, Olaf Hering wrote:

>Am 16.10.2015 um 11:49 schrieb Stephan Kulow:
>> Not that easy dude - e.g. you need automake for some of the above. But
>> if you have a bot ready to handle such cases, I would love to enable it
>> for Leap actually. I hate to wait for i586 libreoffice ;(
>
>Not sure how to automate it.

automake, not automate :)

To build a baselibs.conf-having something.i586 package, its
BuildRequires also need to be present in the i586 scheduler space, also
because .noarch.rpm is not normally copied between schedulers.
--
To unsubscribe, e-mail: [hidden email]
To contact the owner, e-mail: [hidden email]

Reply | Threaded
Open this post in threaded view
|

Re: We'll stop testing i586 for TW

Olaf Hering-2
On Fri, Oct 16, Jan Engelhardt wrote:

> automake, not automate :)

Sure automate, who would want to click through more than a dozen pkgs to
disable building?

> To build a baselibs.conf-having something.i586 package, its
> BuildRequires also need to be present in the i586 scheduler space, also
> because .noarch.rpm is not normally copied between schedulers.

This could be caught by looking for state "unresolvable", after wiping
everything with build-disabled. After all the whole thing needs to be
done just once I think.

Regarding the globally enable/disable state, I think globally disable is
better because branching a pkg will most likely inherit this setting in
the branched prj.

Olaf
--
To unsubscribe, e-mail: [hidden email]
To contact the owner, e-mail: [hidden email]

Reply | Threaded
Open this post in threaded view
|

Re: We'll stop testing i586 for TW

Michal Marek
In reply to this post by Olaf Hering-2
On 2015-10-16 11:56, Olaf Hering wrote:

> Am 16.10.2015 um 11:49 schrieb Stephan Kulow:
>> Not that easy dude - e.g. you need automake for some of the above. But
>> if you have a bot ready to handle such cases, I would love to enable it
>> for Leap actually. I hate to wait for i586 libreoffice ;(
>
> I did not mean to do it now, just when the time has come.
>
> Not sure how to automate it. Either x86 gets globally disabled, and the
> base chroot + each pkg with baselibs gets enabled. Or x86 remains
> globablly enabled, and each pkg without baselibs - base chroot gets
> disabled.
>
> Shouldnt it be like "obs ls -e $prj $pkg baselibs.conf && enable()"?

I'm trying this right now:

osc ls openSUSE:Factory | sort -u >all-pkgs
for p in $(< all-pkgs); do
        if osc ls openSUSE:Factory "$p" | grep -Fqx baselibs.conf; then
                echo "$p"
        fi
done >baselibs-pkgs
for p in $(< baselibs-pkgs); do
        printf '%s\n' $(osc dependson openSUSE:Factory "$p" standard i586 | sed
's/://')
done | sort -u >keep-pkgs
join -v1 all-pkgs keep-pkgs

It's still creating the baselibs list though, and it's not handling
errors from osc. But I'm curious what is going to be the diff.

Michal
--
To unsubscribe, e-mail: [hidden email]
To contact the owner, e-mail: [hidden email]

Reply | Threaded
Open this post in threaded view
|

Re: We'll stop testing i586 for TW

T. Godawa
In reply to this post by Stephan Kulow-3
Hi Stephan,

> As we are under heavy load on openqa.opensuse.org with 2 releases and
> its stagings, we'll stop testing the i586 ISOs for TW. I won't change
> anything about them being built for now, but the next step would be
> to publish these ISOs somewhere outside of /tumbleweed/iso.
well, not good, no 32bit-version for Leap and in near future none for TW.

> I'm not so sure about the timeline we should follow with this, but as
> you might know by now, my interest in historical platforms isn't high.
I know and for most scenarios that is ok. But not all people using Linux
outside the "developer world" are using latest and greatest hardware.

And a ThinkPad with 2 GHz CoreDuo, 4 GB of RAM and a small SSD is still
a reliable and very cheap "working horse" for many people, when using it
with Linux!
--

Kind regards,

Thorolf
--
To unsubscribe, e-mail: [hidden email]
To contact the owner, e-mail: [hidden email]

Reply | Threaded
Open this post in threaded view
|

Re: We'll stop testing i586 for TW

Stephan Kulow-3
Am 17.10.2015 um 23:24 schrieb T. Godawa:

> And a ThinkPad with 2 GHz CoreDuo, 4 GB of RAM and a small SSD is still
> a reliable and very cheap "working horse" for many people, when using it
> with Linux!
>
I'm sure it is. And as I said: as long as these many people fix their
own bugs I'm fine with keeping it around as a port somewhere.

Greetings, Stephan

--
Ma muaß weiterkämpfen, kämpfen bis zum Umfalln, a wenn die
ganze Welt an Arsch offen hat, oder grad deswegn.
--
To unsubscribe, e-mail: [hidden email]
To contact the owner, e-mail: [hidden email]

Reply | Threaded
Open this post in threaded view
|

Re: We'll stop testing i586 for TW

Christian Boltz-5
Hello,

Am Sonntag, 18. Oktober 2015 schrieb Stephan Kulow:
> Am 17.10.2015 um 23:24 schrieb T. Godawa:
> > And a ThinkPad with 2 GHz CoreDuo, 4 GB of RAM and a small SSD is
> > still a reliable and very cheap "working horse" for many people,
> > when using it with Linux!
>
> I'm sure it is. And as I said: as long as these many people fix their
> own bugs I'm fine with keeping it around as a port somewhere.

To make sure that i586-specific bugs get noticed quickly, could you
re-enable one or two i586 tests in openQA?

I'd expect that arch-related failures typically happen on the very base
of the system (booting, kernel, glibc) and not on the upper level (KDE,
whatever) [1], so having one or two i586 tests active sounds like an
acceptable tradeoff between openQA load and keeping i586 supported.

It probably doesn't matter which tests will run, so feel free to pick a
fast test (maybe installation and some xfce testing?) ;-)


Regards,

Christian Boltz

[1] at least that's my guess ;-)
--
> Gibt es eine CPU Beschränkung bei der Prof. Version?
Die gibt es tatsaechlich, hat aber nichts mit der Professional Version
zu tun, sondern mit dem Linux-Kernel selbst. Das Limit liegt aber weit
jenseits von dem, was für Dich vermutlich relevant und bezahlbar ist ;-)
[> Robert und Thomas Hertweck in suse-linux]

--
To unsubscribe, e-mail: [hidden email]
To contact the owner, e-mail: [hidden email]