Re: [oi-dev] Git dependency to

[Aurélien Larcher]

On Fri, Sep 28, 2018 at 1:50 PM Alexander Pyhalov via oi-dev <
oi-dev@openindiana.org> wrote:

> Hi.
> It's part of g11n. These are illumos-specific iconv libraries for
> converting files between characterr sets. Something should go very wrong
> for git to get dependency on them.
>

Thanks for the hint Alexander.
Until the cause is found I will ignore the git component.

BTW, where are old g11n repositories now that hg.openindiana.org has gone?




>
> С уважением,
> Александр Пыхалов,
> программист отдела телекоммуникационной инфраструктуры
> управления информационно-коммуникационной инфраструктуры ЮФУ
>
>
> 
> От: Aurélien Larcher 
> Отправлено: 28 сентября 2018 г. 14:00:19
> Кому: OpenIndiana Developer mailing list
> Тема: [oi-dev] Git dependency to
>
> Hi,
> latest git component failed to publish in the gcc-8 build zone since it
> depends on /system/library/iconv/extra.
>
> A quick check shows that
>
> system/library/iconv/extra@0.5.11,5.11-2013.0.0.0:20151027T075807Z<
> http://pkg.openindiana.org/hipster/info/0/system%2Flibrary%2Ficonv%2Fextra%400.5.11%2C5.11-2013.0.0.0%3A20151027T075807Z
> >
>
> is one of those package that have not been migrated to oi-userland.
>
> Any idea where it is located or whether this dependency is actually
> required at all?
> The only missing file in the workspace is
> '/usr/share/man/man1/git-commit-graph.1'
>
> Aurélien
> --
> ---
> Praise the Caffeine embeddings
> ___
> oi-dev mailing list
> oi-dev@openindiana.org
> https://openindiana.org/mailman/listinfo/oi-dev



-- 
---
Praise the Caffeine embeddings
___
oi-dev mailing list
oi-dev@openindiana.org
https://openindiana.org/mailman/listinfo/oi-dev

Re: [oi-dev] Whitespace in Makefiles in oi-userland keep causing Problems

[Aurélien Larcher]

On Fri, Sep 28, 2018 at 1:48 PM Till Wegmüller  wrote:

> I managed to get it Working by hardcoding every Value with a Whitespace
> inside it in the Package manifest.
>
> Looks like the problem seems to be the way gmake passes Variables to
> pkgmogrify. It usualy does so by specifying -D $VARIABLE on the command
> line. That seems not to be working consistently.
>
> So lesson from that seems to be that we need to hardcode values with
> whitespace into the manifest.
>
> Thanks for the pointers.
>

I am not sure this is the case, there are other instances of values with
whitespaces.
The way values are quoted should be influencial depending on how the recipe
passes them to pkgmogrify.
There is a similar problem with non-escaping dots in PKG_MACROS.



>
> Greetings
> Till
>
> On 09/28/18 12:23 PM, Joerg Schilling wrote:
> > Till Wegmüller  wrote:
> >
> >> A Makefile like so:
> >>
> >> 
> >> COMPONENT_NAME=salt
> >> COMPONENT_VERSION=2018.3.2
> >> COMPONENT_SUMMARY='Portable, distributed, remote execution and
> >> configuration management system'
> >> COMPONENT_PROJECT_URL=http://saltstack.com
> >> COMPONENT_FMRI=system/management/$(COMPONENT_NAME)
> >> COMPONENT_CLASSIFICATION='System/Administration and Configuration'
> >> 
> >>
> >> Something like COMPONENT_CLASSIFICATION would break pkgmogrify with:
> >> pkgmogrify: File
> >>
> /export/home/toast/workspace/openindiana/userland/components/python/salt/build/manifest-i386-salt-27.p5m
> >> line 18: Malformed action at position: 90: whitespace in key:
> >> set name=info.classification
> >> value=org.opensolaris.category.2008:System/Administration and
> Configuration
> >
> > It would help if you could investigate and report on what happenes that
> causes
> > your problem and whether you are unhappy with "make" or with the called
> > program...
> >
> > BTW: do you use "gmake" or make?
> > gmake is known to have a buggy white space handling.
> >
> > Jörg
> >
>
> ___
> oi-dev mailing list
> oi-dev@openindiana.org
> https://openindiana.org/mailman/listinfo/oi-dev



-- 
---
Praise the Caffeine embeddings
___
oi-dev mailing list
oi-dev@openindiana.org
https://openindiana.org/mailman/listinfo/oi-dev

Re: [oi-dev] Git dependency to

[Alexander Pyhalov via oi-dev]

Hi.
It's part of g11n. These are illumos-specific iconv libraries for converting 
files between characterr sets. Something should go very wrong for git to get 
dependency on them.

С уважением,
Александр Пыхалов,
программист отдела телекоммуникационной инфраструктуры
управления информационно-коммуникационной инфраструктуры ЮФУ



От: Aurélien Larcher 
Отправлено: 28 сентября 2018 г. 14:00:19
Кому: OpenIndiana Developer mailing list
Тема: [oi-dev] Git dependency to

Hi,
latest git component failed to publish in the gcc-8 build zone since it depends 
on /system/library/iconv/extra.

A quick check shows that

system/library/iconv/extra@0.5.11,5.11-2013.0.0.0:20151027T075807Z

is one of those package that have not been migrated to oi-userland.

Any idea where it is located or whether this dependency is actually required at 
all?
The only missing file in the workspace is 
'/usr/share/man/man1/git-commit-graph.1'

Aurélien
--
---
Praise the Caffeine embeddings
___
oi-dev mailing list
oi-dev@openindiana.org
https://openindiana.org/mailman/listinfo/oi-dev

Re: [oi-dev] Whitespace in Makefiles in oi-userland keep causing Problems

[Till Wegmüller]

I managed to get it Working by hardcoding every Value with a Whitespace
inside it in the Package manifest.

Looks like the problem seems to be the way gmake passes Variables to
pkgmogrify. It usualy does so by specifying -D $VARIABLE on the command
line. That seems not to be working consistently.

So lesson from that seems to be that we need to hardcode values with
whitespace into the manifest.

Thanks for the pointers.

Greetings
Till

On 09/28/18 12:23 PM, Joerg Schilling wrote:
> Till Wegmüller  wrote:
> 
>> A Makefile like so:
>>
>> 
>> COMPONENT_NAME=salt
>> COMPONENT_VERSION=2018.3.2
>> COMPONENT_SUMMARY='Portable, distributed, remote execution and
>> configuration management system'
>> COMPONENT_PROJECT_URL=http://saltstack.com
>> COMPONENT_FMRI=system/management/$(COMPONENT_NAME)
>> COMPONENT_CLASSIFICATION='System/Administration and Configuration'
>> 
>>
>> Something like COMPONENT_CLASSIFICATION would break pkgmogrify with:
>> pkgmogrify: File
>> /export/home/toast/workspace/openindiana/userland/components/python/salt/build/manifest-i386-salt-27.p5m
>> line 18: Malformed action at position: 90: whitespace in key:
>> set name=info.classification
>> value=org.opensolaris.category.2008:System/Administration and Configuration
> 
> It would help if you could investigate and report on what happenes that causes
> your problem and whether you are unhappy with "make" or with the called 
> program... 
> 
> BTW: do you use "gmake" or make?
> gmake is known to have a buggy white space handling.
> 
> Jörg
> 

___
oi-dev mailing list
oi-dev@openindiana.org
https://openindiana.org/mailman/listinfo/oi-dev

[oi-dev] Git dependency to

[Aurélien Larcher]

Hi,
latest git component failed to publish in the gcc-8 build zone since it
depends on /system/library/iconv/extra.

A quick check shows that

system/library/iconv/extra@0.5.11,5.11-2013.0.0.0:20151027T075807Z


is one of those package that have not been migrated to oi-userland.

Any idea where it is located or whether this dependency is actually
required at all?
The only missing file in the workspace is
'/usr/share/man/man1/git-commit-graph.1'

Aurélien
-- 
---
Praise the Caffeine embeddings
___
oi-dev mailing list
oi-dev@openindiana.org
https://openindiana.org/mailman/listinfo/oi-dev

Re: [oi-dev] Whitespace in Makefiles in oi-userland keep causing Problems

[Joerg Schilling]

Reginald Beardsley via oi-dev  wrote:

> Whitespace in Makefiles has been a mess from the start.  The story is that 
> Feldman realized that making distinctions between types of white space was a 
> bad idea.  But he had a dozen users and didn't want to break their Makefiles. 
>  And the Bourne shell is not much better on that score.  What you can't see 
> can have devastating consequences,

Looks like you missunderstood the concept.

Some people have been unhappy with the concept that action lines need to start 
with an intial TAB.

White space in variables is well defined:

-   initial whitespace is skipped

-   trailing whitespace is removed by SunPro Make but kept by gmake and 
smake, so be careful with # comment at the end of a variable 
assignement line.

-   white space in the middle is required by POSIX to be kept intact
but gmake e.g. shrinks multiple white space to a single white space

The reported problem does not seem to be related to these white space problems.

Jörg

-- 
 EMail:jo...@schily.net(home) Jörg Schilling D-13353 Berlin
joerg.schill...@fokus.fraunhofer.de (work) Blog: http://schily.blogspot.com/
 URL: http://cdrecord.org/private/ http://sf.net/projects/schilytools/files/'

___
oi-dev mailing list
oi-dev@openindiana.org
https://openindiana.org/mailman/listinfo/oi-dev

Re: [oi-dev] 2018.10 targets

[Aurélien Larcher]

On Fri, Sep 28, 2018 at 10:41 AM Alexander Pyhalov via oi-dev <
oi-dev@openindiana.org> wrote:

> Hi, folks.
>
> We are approaching October.
> While I have limited time now, I want to coordinate priorities and
> targets, what should we do till end of October (and due to personal
> reasons, I'd better announce snapshot till 20 October or earlier or allow
> someone else to do this).
>
> We have several regressions and problems, which would be good to fix.
>
> 1) Virtualbox failure after FPU changes
> https://www.illumos.org/issues/9761 ,
> https://www.virtualbox.org/ticket/17947 is a serious one, but I'm afraid
> I lack skills to fix it in reasonable time. The stack trace itself doesn't
> seem related to illumos-gate changes, but likely it is. In KVM Joyent fixed
> this using new hma_fpu_* API, but Virtualbox structures to store VM
> registers differ and from brief overlook it has several asm functions to
> store/restore Host/Base registers state. So far I don't have an idea how
> convert one to another. So, this likely will be broken in 2018.10 unless
> someone skillful dig into this.
>
> 2) sbcl issues - threaded sbcl can not be built after KPTI changes, it
> fails FLOCK and posix tests. Needs investigation. It's a serious issue,
> because threaded sbcl is needed to build pgloader.
>
> 3) Our Mate is a bit old, but newer Mate requires updated GTK3 .  Updated
> GTK3 IIRC has dropped support for theme engines, and so our Nimbus theme
> based on unico engine doesn't work. Also I've heard there were issues with
> new mate-terminal.
> Michal, do you have some comments here?
>
> 4) As for new features, I really want to see KVM zones, perhaps, I'll dig
> into this shortly.
>
> Does somebody have another targets/comments?
>
> Please, avoid turning this topic in bike-shedding, I want to hear OI
> developers in a way "I ll take item #N" or "I expect to finish feature A
> before snapshot", not "please, also fix this and that".
>

Hi Alexander,
I'll be busy as well and my skills are not up to tasks #1 or #2... the only
tracks I can pursue for now are:
1) fixing pkg and gfx-drm components so that we can migrate userland to
gcc-8 before or after the snapshot.
2) pushing Firefox 60esr
3) updating Xorg to latest
4) fixing openmpi which has been broken before summer

If I have time to fix audacity then we can also update ffmpeg and vlc.
I had some fixes for drm/i915 but my Intel box died 4 months ago so this is
on hold until I can replace it.
Kind regards,

Aurélien


>
> Best regards,
> Alexander Pyhalov,
> system administrator of Southern Federal University IT department
>
>
>
> ___
> oi-dev mailing list
> oi-dev@openindiana.org
> https://openindiana.org/mailman/listinfo/oi-dev
>


-- 
---
Praise the Caffeine embeddings
___
oi-dev mailing list
oi-dev@openindiana.org
https://openindiana.org/mailman/listinfo/oi-dev

Re: [oi-dev] 2018.10 targets

[Till Wegmüller]

Hi

My personal target would be to have saltstack packaged and some
tutorials written how to use it.

As for KVM Zones. If we modify IPKG Brand and inject the appropriate
permissions and devices it should work. Although I would prefer it being
a fork of Tribblixes Sparse Brand. Also have a look into kvmadm what it
does to prepare KVM in zone setup. It has that capability but i did not
manage to make it work.

Greetings
Till

On 09/28/18 10:41 AM, Alexander Pyhalov via oi-dev wrote:
> Hi, folks.
> 
> We are approaching October.
> While I have limited time now, I want to coordinate priorities and targets, 
> what should we do till end of October (and due to personal reasons, I'd 
> better announce snapshot till 20 October or earlier or allow someone else to 
> do this).
> 
> We have several regressions and problems, which would be good to fix.
> 
> 1) Virtualbox failure after FPU changes https://www.illumos.org/issues/9761 , 
> https://www.virtualbox.org/ticket/17947 is a serious one, but I'm afraid I 
> lack skills to fix it in reasonable time. The stack trace itself doesn't seem 
> related to illumos-gate changes, but likely it is. In KVM Joyent fixed this 
> using new hma_fpu_* API, but Virtualbox structures to store VM registers 
> differ and from brief overlook it has several asm functions to store/restore 
> Host/Base registers state. So far I don't have an idea how convert one to 
> another. So, this likely will be broken in 2018.10 unless someone skillful 
> dig into this.
> 
> 2) sbcl issues - threaded sbcl can not be built after KPTI changes, it fails 
> FLOCK and posix tests. Needs investigation. It's a serious issue, because 
> threaded sbcl is needed to build pgloader.
> 
> 3) Our Mate is a bit old, but newer Mate requires updated GTK3 .  Updated 
> GTK3 IIRC has dropped support for theme engines, and so our Nimbus theme 
> based on unico engine doesn't work. Also I've heard there were issues with 
> new mate-terminal. 
> Michal, do you have some comments here?
> 
> 4) As for new features, I really want to see KVM zones, perhaps, I'll dig 
> into this shortly.
> 
> Does somebody have another targets/comments? 
> 
> Please, avoid turning this topic in bike-shedding, I want to hear OI 
> developers in a way "I ll take item #N" or "I expect to finish feature A 
> before snapshot", not "please, also fix this and that". 
> 
> Best regards, 
> Alexander Pyhalov, 
> system administrator of Southern Federal University IT department
> 
> 
> 
> ___
> oi-dev mailing list
> oi-dev@openindiana.org
> https://openindiana.org/mailman/listinfo/oi-dev
> 

___
oi-dev mailing list
oi-dev@openindiana.org
https://openindiana.org/mailman/listinfo/oi-dev

Re: [oi-dev] Whitespace in Makefiles in oi-userland keep causing Problems

[Till Wegmüller]

No Unfortunately not...
And yes it seems to be a complete mess with whitespaces

And it is pkgmogrify which complains. not Make itself

On 09/28/18 10:03 AM, Alexander Pyhalov via oi-dev wrote:
> Does
> COMPONENT_CLASSIFICATION=   System/Administration and Configuration
> work?
> 
> С уважением,
> Александр Пыхалов,
> программист отдела телекоммуникационной инфраструктуры
> управления информационно-коммуникационной инфраструктуры ЮФУ
> 
> 
> 
> От: Till Wegmüller 
> Отправлено: 27 сентября 2018 г. 22:03:14
> Кому: OpenIndiana Developer mailing list
> Тема: [oi-dev] Whitespace in Makefiles in oi-userland keep causing  
> Problems
> 
> Hello Fellow developers.
> 
> I have a confusing Problem with our buildsystem.
> Since today it keeps behaving Inconsistently when I have a Whitespace in
> a Value of a variable.
> 
> A Makefile like so:
> 
> 
> COMPONENT_NAME=salt
> COMPONENT_VERSION=2018.3.2
> COMPONENT_SUMMARY='Portable, distributed, remote execution and
> configuration management system'
> COMPONENT_PROJECT_URL=http://saltstack.com
> COMPONENT_FMRI=system/management/$(COMPONENT_NAME)
> COMPONENT_CLASSIFICATION='System/Administration and Configuration'
> 
> 
> Something like COMPONENT_CLASSIFICATION would break pkgmogrify with:
> pkgmogrify: File
> /export/home/toast/workspace/openindiana/userland/components/python/salt/build/manifest-i386-salt-27.p5m
> line 18: Malformed action at position: 90: whitespace in key:
> set name=info.classification
> value=org.opensolaris.category.2008:System/Administration and Configuration
> 
> I could manage to supress this by using escape characters like " or '.
> But not in all cases. And with the Classification the Whitespaces are
> required as that Category has Whitespaces in their name. Thus it won't
> publish.
> 
> I remember that Whitespaces where no problem not long ago. How will we
> handle this? Packages like puppet and many others have also Whitespaces
> in the Makefiles.
> 
> Greetings
> Till
> 
> ___
> oi-dev mailing list
> oi-dev@openindiana.org
> https://openindiana.org/mailman/listinfo/oi-dev
> ___
> oi-dev mailing list
> oi-dev@openindiana.org
> https://openindiana.org/mailman/listinfo/oi-dev
> 

___
oi-dev mailing list
oi-dev@openindiana.org
https://openindiana.org/mailman/listinfo/oi-dev

[oi-dev] 2018.10 targets

[Alexander Pyhalov via oi-dev]

Hi, folks.

We are approaching October.
While I have limited time now, I want to coordinate priorities and targets, 
what should we do till end of October (and due to personal reasons, I'd better 
announce snapshot till 20 October or earlier or allow someone else to do this).

We have several regressions and problems, which would be good to fix.

1) Virtualbox failure after FPU changes https://www.illumos.org/issues/9761 , 
https://www.virtualbox.org/ticket/17947 is a serious one, but I'm afraid I lack 
skills to fix it in reasonable time. The stack trace itself doesn't seem 
related to illumos-gate changes, but likely it is. In KVM Joyent fixed this 
using new hma_fpu_* API, but Virtualbox structures to store VM registers differ 
and from brief overlook it has several asm functions to store/restore Host/Base 
registers state. So far I don't have an idea how convert one to another. So, 
this likely will be broken in 2018.10 unless someone skillful dig into this.

2) sbcl issues - threaded sbcl can not be built after KPTI changes, it fails 
FLOCK and posix tests. Needs investigation. It's a serious issue, because 
threaded sbcl is needed to build pgloader.

3) Our Mate is a bit old, but newer Mate requires updated GTK3 .  Updated GTK3 
IIRC has dropped support for theme engines, and so our Nimbus theme based on 
unico engine doesn't work. Also I've heard there were issues with new 
mate-terminal. 
Michal, do you have some comments here?

4) As for new features, I really want to see KVM zones, perhaps, I'll dig into 
this shortly.

Does somebody have another targets/comments? 

Please, avoid turning this topic in bike-shedding, I want to hear OI developers 
in a way "I ll take item #N" or "I expect to finish feature A before snapshot", 
not "please, also fix this and that". 

Best regards, 
Alexander Pyhalov, 
system administrator of Southern Federal University IT department



___
oi-dev mailing list
oi-dev@openindiana.org
https://openindiana.org/mailman/listinfo/oi-dev

Re: [oi-dev] Whitespace in Makefiles in oi-userland keep causing Problems

[Alexander Pyhalov via oi-dev]

Does
COMPONENT_CLASSIFICATION=   System/Administration and Configuration
work?

С уважением,
Александр Пыхалов,
программист отдела телекоммуникационной инфраструктуры
управления информационно-коммуникационной инфраструктуры ЮФУ



От: Till Wegmüller 
Отправлено: 27 сентября 2018 г. 22:03:14
Кому: OpenIndiana Developer mailing list
Тема: [oi-dev] Whitespace in Makefiles in oi-userland keep causing  Problems

Hello Fellow developers.

I have a confusing Problem with our buildsystem.
Since today it keeps behaving Inconsistently when I have a Whitespace in
a Value of a variable.

A Makefile like so:


COMPONENT_NAME=salt
COMPONENT_VERSION=2018.3.2
COMPONENT_SUMMARY='Portable, distributed, remote execution and
configuration management system'
COMPONENT_PROJECT_URL=http://saltstack.com
COMPONENT_FMRI=system/management/$(COMPONENT_NAME)
COMPONENT_CLASSIFICATION='System/Administration and Configuration'


Something like COMPONENT_CLASSIFICATION would break pkgmogrify with:
pkgmogrify: File
/export/home/toast/workspace/openindiana/userland/components/python/salt/build/manifest-i386-salt-27.p5m
line 18: Malformed action at position: 90: whitespace in key:
set name=info.classification
value=org.opensolaris.category.2008:System/Administration and Configuration

I could manage to supress this by using escape characters like " or '.
But not in all cases. And with the Classification the Whitespaces are
required as that Category has Whitespaces in their name. Thus it won't
publish.

I remember that Whitespaces where no problem not long ago. How will we
handle this? Packages like puppet and many others have also Whitespaces
in the Makefiles.

Greetings
Till

___
oi-dev mailing list
oi-dev@openindiana.org
https://openindiana.org/mailman/listinfo/oi-dev
___
oi-dev mailing list
oi-dev@openindiana.org
https://openindiana.org/mailman/listinfo/oi-dev