[Xenomai] about Xenomai in Debian and Ubuntu

[Xenomai] about Xenomai in Debian and Ubuntu

[Tom Z]

Hi,

I noticed that Debian and Ubuntu desktop distro's have included Xenomai, 
and I can "install" Xenomai through the software center in these 
distro's. But after I "installed" Xenomai in Ubuntu, executing  "dmesg | grep xenomai" showed nothing, so I am wondering if I still need to 
patch and compile the kernels before using the Xenomai features? If so, 
what benefits does it have to include Xenomai in these distro's.

Thanks,
Tom

Re: [Xenomai] about Xenomai in Debian and Ubuntu

[Roland Stigge]

Hi,

On 06/03/13 04:36, Tom Z wrote:
> I noticed that Debian and Ubuntu desktop distro's have included Xenomai, 
> and I can "install" Xenomai through the software center in these 
> distro's.

I can only speak for Debian here. We deliberately don't promote it as
"desktop distro". Rather as "Universal OS", to underline its use in the
Server, Embedded, Realtime areas, etc.

> But after I "installed" Xenomai in Ubuntu, executing  "dmesg | grep xenomai" showed nothing, so I am wondering if I still need to 
> patch and compile the kernels before using the Xenomai features? If so, 
> what benefits does it have to include Xenomai in these distro's.

First, you still have to compile the Xenomai-enabled kernel yourself
(but this might change in the future). This is supported by the
Debian-included package linux-patch-xenomai. The file

/usr/share/doc/linux-patch-xenomai/README.Debian

gives instructions how to prepare the kernel as a Debian package.

Second, Debian ships Xenomai's userland programs and libraries, so you
can use them directly, without compiling.

Since Debian derived distros (as you mentioned one) typically copy the
Xenomai packages from Debian, much of this should apply there also, but
I can't guarantee this.

HTH,

Roland

Re: [Xenomai] about Xenomai in Debian and Ubuntu

[Gilles Chanteperdrix]

On 03/06/2013 08:43 AM, Roland Stigge wrote:

> Hi,
> 
> On 06/03/13 04:36, Tom Z wrote:
>> I noticed that Debian and Ubuntu desktop distro's have included Xenomai, 
>> and I can "install" Xenomai through the software center in these 
>> distro's.
> 
> I can only speak for Debian here. We deliberately don't promote it as
> "desktop distro". Rather as "Universal OS", to underline its use in the
> Server, Embedded, Realtime areas, etc.
> 
>> But after I "installed" Xenomai in Ubuntu, executing  "dmesg | grep xenomai" showed nothing, so I am wondering if I still need to 
>> patch and compile the kernels before using the Xenomai features? If so, 
>> what benefits does it have to include Xenomai in these distro's.
> 
> First, you still have to compile the Xenomai-enabled kernel yourself
> (but this might change in the future). This is supported by the
> Debian-included package linux-patch-xenomai. The file
> 
> /usr/share/doc/linux-patch-xenomai/README.Debian
> 
> gives instructions how to prepare the kernel as a Debian package.
> 
> Second, Debian ships Xenomai's userland programs and libraries, so you
> can use them directly, without compiling.
> 
> Since Debian derived distros (as you mentioned one) typically copy the
> Xenomai packages from Debian, much of this should apply there also, but
> I can't guarantee this.


As a side note, we started with 2.6.2.1 to provide pre-compiled kernel
packages for the Debian stable distribution, simply add:

deb http://www.xenomai.org/debian squeeze main

Then run:
apt-get update
apt-get install xenomai-keyring
Answer yes
apt-get update
apt-get install linux-image-3.5.7-xenomai-2.6.2.1 xenomai-runtime

Then reboot on the new kernel, and you should be able to launch:
xeno latency

The Linux CNC project also provides packages for Ubuntu Precise, Debian
Squeeze and Ubuntu Lucid, see:

http://wiki.linuxcnc.org/cgi-bin/wiki.pl?XenomaiKernelPackages

-- 
                                                                Gilles.

Re: [Xenomai] about Xenomai in Debian and Ubuntu

[Gilles Chanteperdrix]

On 03/06/2013 08:43 AM, Roland Stigge wrote:

> Hi,
> 
> On 06/03/13 04:36, Tom Z wrote:
>> I noticed that Debian and Ubuntu desktop distro's have included Xenomai, 
>> and I can "install" Xenomai through the software center in these 
>> distro's.
> 
> I can only speak for Debian here. We deliberately don't promote it as
> "desktop distro". Rather as "Universal OS", to underline its use in the
> Server, Embedded, Realtime areas, etc.
> 
>> But after I "installed" Xenomai in Ubuntu, executing  "dmesg | grep xenomai" showed nothing, so I am wondering if I still need to 
>> patch and compile the kernels before using the Xenomai features? If so, 
>> what benefits does it have to include Xenomai in these distro's.
> 
> First, you still have to compile the Xenomai-enabled kernel yourself
> (but this might change in the future).


Roland, at some point I lost track of what your intentions were. In the
following mail:
http://www.xenomai.org/pipermail/xenomai/2012-July/000473.html

You seemed to agree to take care of the debian directory in xenomai
sources, asking me to simply warn you before a release.
Which I did before the 2.6.2 release in the following mail:

http://www.xenomai.org/pipermail/xenomai/2012-December/027051.html

With a list of things which could be changed in the debian/rules file,
inviting you to do what I had understood you had agreed to do.

In this answer:
http://www.xenomai.org/pipermail/xenomai/2012-December/027052.html
You simply told me that you did not even try debian/rules, following
this, Leopold sent a post with a patch to fix the issue before the
release, but your reply could be interpreted as the fact that the patch
was actually broken, you did not send a fixed patch for the update, and
as a result 2.6.2 was released with a broken debian directory.

Which is why, for the 2.6.2.1 release, I did not warn you (since it had
proved useless anyway) and took the liberty to do the changes to
debian/rules myself, as, according to what I had said many times on this
list, I started to set-up a debian repository (following your advices on
the way to do it, use reprepro), and so needed a working debian/rules.
The aim of this was to avoid the kind of questions "Tom Z" is asking.

Now, since I did the Debian packages following your advice, and
basically did the work that you did not want to do, I do not really
understand why you keep ignoring the fact that we have not one but two
sources of up to date pre-buid debian kernel packages for Debian
squeeze, and continue to give answers to "Tom Z" misguiding him in
compiling the Xenomai package from the tarball, not even from the debian
directory.

Regards.


-- 
                                                                Gilles.

Re: [Xenomai] about Xenomai in Debian and Ubuntu

[Roland Stigge]

On 03/12/2013 09:57 PM, Gilles Chanteperdrix wrote:
> Roland, at some point I lost track of what your intentions were. In the
> following mail:
> http://www.xenomai.org/pipermail/xenomai/2012-July/000473.html
> 
> You seemed to agree to take care of the debian directory in xenomai
> sources, asking me to simply warn you before a release.
> Which I did before the 2.6.2 release in the following mail:
> 
> http://www.xenomai.org/pipermail/xenomai/2012-December/027051.html
> 
> With a list of things which could be changed in the debian/rules file,
> inviting you to do what I had understood you had agreed to do.

When you propose things, and implement them, fine. That's your decision.
When you propose things and I implement them, also fine. That's my
decision. But please don't expect me to just follow commands. That's not
my nature at all. Sorry.

> http://www.xenomai.org/pipermail/xenomai/2012-December/027052.html
> You simply told me that you did not even try debian/rules, following
> this, Leopold sent a post with a patch to fix the issue before the
> release, but your reply could be interpreted as the fact that the patch
> was actually broken, you did not send a fixed patch for the update, and
> as a result 2.6.2 was released with a broken debian directory.

Please don't over-interpret things I actually didn't say at all. Every
patch is welcome, including things you propose, and from other
contributors. When I see conflicts in the code, I will state that clearly.

> Which is why, for the 2.6.2.1 release, I did not warn you (since it had
> proved useless anyway)

This is also your interpretation of things, and your own decision. Not
discussed with me.

Sorry when I missed some detail in the past. Please don't let this lead
to bad further communication in return, or blaming each other.

Thanks,

Roland


PS: I consider Debian itself the primary source of Debian packages and
will continue to do so. When people ask specific questions about those,
I'll answer accordingly.

PPS: If you consider your tone as helpful towards positive motivation of
contributors, please rethink.

Re: [Xenomai] about Xenomai in Debian and Ubuntu

[Stéphane LOS]

Le 12/03/2013 22:28, Roland Stigge a écrit :
> On 03/12/2013 09:57 PM, Gilles Chanteperdrix wrote:
>> Roland, at some point I lost track of what your intentions were. In the
>> following mail:
>> http://www.xenomai.org/pipermail/xenomai/2012-July/000473.html
>>
>> You seemed to agree to take care of the debian directory in xenomai
>> sources, asking me to simply warn you before a release.
>> Which I did before the 2.6.2 release in the following mail:
>>
>> http://www.xenomai.org/pipermail/xenomai/2012-December/027051.html
>>
>> With a list of things which could be changed in the debian/rules file,
>> inviting you to do what I had understood you had agreed to do.
> When you propose things, and implement them, fine. That's your decision.
> When you propose things and I implement them, also fine. That's my
> decision. But please don't expect me to just follow commands. That's not
> my nature at all. Sorry.
>
>> http://www.xenomai.org/pipermail/xenomai/2012-December/027052.html
>> You simply told me that you did not even try debian/rules, following
>> this, Leopold sent a post with a patch to fix the issue before the
>> release, but your reply could be interpreted as the fact that the patch
>> was actually broken, you did not send a fixed patch for the update, and
>> as a result 2.6.2 was released with a broken debian directory.
> Please don't over-interpret things I actually didn't say at all. Every
> patch is welcome, including things you propose, and from other
> contributors. When I see conflicts in the code, I will state that clearly.
>
>> Which is why, for the 2.6.2.1 release, I did not warn you (since it had
>> proved useless anyway)
> This is also your interpretation of things, and your own decision. Not
> discussed with me.
>
> Sorry when I missed some detail in the past. Please don't let this lead
> to bad further communication in return, or blaming each other.
>
> Thanks,
>
> Roland
>
>
> PS: I consider Debian itself the primary source of Debian packages and
> will continue to do so. When people ask specific questions about those,
> I'll answer accordingly.
>
> PPS: If you consider your tone as helpful towards positive motivation of
> contributors, please rethink.
>
> _______________________________________________
> Xenomai mailing list
> Xenomai@xenomai.org
> http://www.xenomai.org/mailman/listinfo/xenomai
>
Hi,

I just want to say that I agree with Roland about the tone.

My colleague Jérôme and I ARE NOT kernel developers for sure.

Hilscher is proposing a Linux driver for the cifX boards. Not for free, 
I can't do anything about that fact.
Then our customer gets the source code. The documentation is available 
on our web site.

As I like FLOSS and have years of fiddling here and there, I have 
noticed Xenomai existence and usage in projects such as Orocos and Linux 
CNC.
And I have noticed also the usage of CIF boards in those projects.

So I have proposed to my manager and Jérôme to work on a Xenomai driver 
for cifX boards and it was accepted.

Again, we are not kernel developers and our background is different.

Of course we could sell some boards but it is nothing compared to the 
time and energy we have already spent in this project.

Our customers are traditionally using RTX or VxWorks with our boards. I 
wanted to promote the use of Linux / Xenomai for some kind of applications.

We have received some support from Gilles and Jan and I thank them very 
much but the tone was never very friendly.

I understand that FLOSS projects need to sell training and development 
but the cost would have cancelled the project since we don't think we 
could generate the cash to cover the expenses. So the driver is done on 
our free time.

I am just wondering why in this list nobody cares of saying Hi or Hello 
before anything.
May be you are just too busy reading / writing kernel code lines.

I am a little bit disappointed. Sorry for bothering you.

Best Regards,
Cordialement,

Stéphane LOS
slos@hilscher.com
Support technique

~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Hilscher France
12, rue du 35ème Régiment d'Aviation
Miniparc du Chêne
69500 BRON
France
Tél. : +33 (0) 4 72 37 98 40
Fax  : +33 (0) 4 78 26 83 27
http ://www.hilscher.fr
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

Re: [Xenomai] about Xenomai in Debian and Ubuntu

[Jan Kiszka]

Hi Stéphane, (hope that is better ;) )

On 2013-03-13 10:40, Stéphane LOS wrote:
> Hi,
> 
> I just want to say that I agree with Roland about the tone.
> 
> My colleague Jérôme and I ARE NOT kernel developers for sure.
> 
> Hilscher is proposing a Linux driver for the cifX boards. Not for free,
> I can't do anything about that fact.
> Then our customer gets the source code. The documentation is available
> on our web site.
> 
> As I like FLOSS and have years of fiddling here and there, I have
> noticed Xenomai existence and usage in projects such as Orocos and Linux
> CNC.
> And I have noticed also the usage of CIF boards in those projects.
> 
> So I have proposed to my manager and Jérôme to work on a Xenomai driver
> for cifX boards and it was accepted.
> 
> Again, we are not kernel developers and our background is different.
> 
> Of course we could sell some boards but it is nothing compared to the
> time and energy we have already spent in this project.

You should not underestimate the value of code review here. Gilles and I
both pointed out flaws in your code that existed not just due to the
"unusual" target environment. I hope there were some patterns to pick up
for other projects as well.

And don't take it personally, it's human to err. I've seen a lot of code
written by a single person or a certain group that is very "focused",
not taking care of corner cases. And you'll find bugs in my code as well
before - and sometimes unfortunately also after - it went through
thorough review.

> 
> Our customers are traditionally using RTX or VxWorks with our boards. I
> wanted to promote the use of Linux / Xenomai for some kind of applications.

Which is likely a good idea when thinking long-term.

> 
> We have received some support from Gilles and Jan and I thank them very
> much but the tone was never very friendly.
> 
> I understand that FLOSS projects need to sell training and development
> but the cost would have cancelled the project since we don't think we
> could generate the cash to cover the expenses. So the driver is done on
> our free time.

We appreciate your effort, and I'm sorry to hear that you have no
backing to finish this officially - specifically as the result should
provide a better customer experience of your products. I know what you
are talking about, sometimes I'm still involved in similar fights.

> 
> I am just wondering why in this list nobody cares of saying Hi or Hello
> before anything.
> May be you are just too busy reading / writing kernel code lines.
> 
> I am a little bit disappointed. Sorry for bothering you.

The tone may appear rough, sorry, and it is sometimes related to the
number of mails and, thus, contexts we work in every day (various OSS
projects, customer projects, internal stuff...). So some statements
become briefer. Sometimes they also become a bit rougher when requests
to change certain issues are simply ignored without commenting on the why.

Jan

-- 
Siemens AG, Corporate Technology, CT RTC ITP SDP-DE
Corporate Competence Center Embedded Linux

Re: [Xenomai] about Xenomai in Debian and Ubuntu

[Stéphane LOS]

Le 13/03/2013 13:29, Jan Kiszka a écrit :
> Hi Stéphane, (hope that is better ;) )
Hi Jan, (Yes, I feel better ! :-) )
>
> On 2013-03-13 10:40, Stéphane LOS wrote:
>> Hi,
>>
>> I just want to say that I agree with Roland about the tone.
>>
>> My colleague Jérôme and I ARE NOT kernel developers for sure.
>>
>> Hilscher is proposing a Linux driver for the cifX boards. Not for free,
>> I can't do anything about that fact.
>> Then our customer gets the source code. The documentation is available
>> on our web site.
>>
>> As I like FLOSS and have years of fiddling here and there, I have
>> noticed Xenomai existence and usage in projects such as Orocos and Linux
>> CNC.
>> And I have noticed also the usage of CIF boards in those projects.
>>
>> So I have proposed to my manager and Jérôme to work on a Xenomai driver
>> for cifX boards and it was accepted.
>>
>> Again, we are not kernel developers and our background is different.
>>
>> Of course we could sell some boards but it is nothing compared to the
>> time and energy we have already spent in this project.
> You should not underestimate the value of code review here. Gilles and I
> both pointed out flaws in your code that existed not just due to the
> "unusual" target environment. I hope there were some patterns to pick up
> for other projects as well.
We do not underestimate and we try to observe every point you make.
Of course, we know we can learn a lot from Gilles and you.
> And don't take it personally, it's human to err.
I don't take it personally, specially because I am not the developer.
But I feel concerned about Jérôme's feelings.
> I've seen a lot of code
> written by a single person or a certain group that is very "focused",
> not taking care of corner cases. And you'll find bugs in my code as well
> before - and sometimes unfortunately also after - it went through
> thorough review.
You have to consider that we are not kernel coders and we have struggled 
with code coming from UIO, RTDM and Hilscher's toolkit, which we cannot 
offer to review since it is Mister Hilscher property.
We have read the documentation on drivers writing and it has not been 
easy to figure out what was really needed.
So it is likely that there are bugs in our code and we are happy if you 
can help us to get rid of them.
Unfortunately, we can only offer the RTDM code for review and cross 
fingers about the user land stuff.

>
>> Our customers are traditionally using RTX or VxWorks with our boards. I
>> wanted to promote the use of Linux / Xenomai for some kind of applications.
> Which is likely a good idea when thinking long-term.
May be. Everyone has to make a living.
I don't blame anybody who has a different opinion. Hilscher is not an 
Open Source company either.

>
>> We have received some support from Gilles and Jan and I thank them very
>> much but the tone was never very friendly.
>>
>> I understand that FLOSS projects need to sell training and development
>> but the cost would have cancelled the project since we don't think we
>> could generate the cash to cover the expenses. So the driver is done on
>> our free time.
> We appreciate your effort, and I'm sorry to hear that you have no
> backing to finish this officially - specifically as the result should
> provide a better customer experience of your products. I know what you
> are talking about, sometimes I'm still involved in similar fights.
Well, I wish I had a customer buying 500 boards to use with Xenomai. 
This would trigger some interest.

>
>> I am just wondering why in this list nobody cares of saying Hi or Hello
>> before anything.
>> May be you are just too busy reading / writing kernel code lines.
>>
>> I am a little bit disappointed. Sorry for bothering you.
> The tone may appear rough, sorry, and it is sometimes related to the
> number of mails and, thus, contexts we work in every day (various OSS
> projects, customer projects, internal stuff...). So some statements
> become briefer. Sometimes they also become a bit rougher when requests
> to change certain issues are simply ignored without commenting on the why.
Simply ignored, I don't think so. Sometimes one only simply doesn't 
understand what another meant.
We really have no reason to not do what you want us to do.
If we see what you mean, you will get it done.

Gilles and you have a good English level, it is not our case.
We may have other weaknesses.

>
> Jan
>

Best Regards,
Cordialement,

Stéphane LOS
slos@hilscher.com
Support technique

~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Hilscher France
12, rue du 35ème Régiment d'Aviation
Miniparc du Chêne
69500 BRON
France
Tél. : +33 (0) 4 72 37 98 40
Fax  : +33 (0) 4 78 26 83 27
http ://www.hilscher.fr
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

Re: [Xenomai] about Xenomai in Debian and Ubuntu

[Gilles Chanteperdrix]

On 03/13/2013 10:40 AM, Stéphane LOS wrote:

> Hi,


Hi,

I have nothing to add to Jan's answer, except an answer to this specific
remark:

> I understand that FLOSS projects need to sell training and development 
> but the cost would have cancelled the project since we don't think we 
> could generate the cash to cover the expenses. So the driver is done on 
> our free time.


For what it is worth, there is no intention of selling anything when we
do support. Just like you, I answer your questions during my free time.

Regards.

-- 
                                                                Gilles.

Re: [Xenomai] about Xenomai in Debian and Ubuntu

[Gilles Chanteperdrix]

On 03/12/2013 10:28 PM, Roland Stigge wrote:

> On 03/12/2013 09:57 PM, Gilles Chanteperdrix wrote:
>> Roland, at some point I lost track of what your intentions were. In the
>> following mail:
>> http://www.xenomai.org/pipermail/xenomai/2012-July/000473.html
>>
>> You seemed to agree to take care of the debian directory in xenomai
>> sources, asking me to simply warn you before a release.
>> Which I did before the 2.6.2 release in the following mail:
>>
>> http://www.xenomai.org/pipermail/xenomai/2012-December/027051.html
>>
>> With a list of things which could be changed in the debian/rules file,
>> inviting you to do what I had understood you had agreed to do.
> 
> When you propose things, and implement them, fine. That's your decision.
> When you propose things and I implement them, also fine. That's my
> decision. But please don't expect me to just follow commands. That's not
> my nature at all. Sorry.


Actually, what I quoted above is the contrary, you commanded me to warn
you before a release, and I obeyed. The real question at the beginning
of the mail is do you intend or not to maintain the debian directory in
xenomai sources?

> 
>> http://www.xenomai.org/pipermail/xenomai/2012-December/027052.html
>> You simply told me that you did not even try debian/rules, following
>> this, Leopold sent a post with a patch to fix the issue before the
>> release, but your reply could be interpreted as the fact that the patch
>> was actually broken, you did not send a fixed patch for the update, and
>> as a result 2.6.2 was released with a broken debian directory.
> 
> Please don't over-interpret things I actually didn't say at all. Every
> patch is welcome, including things you propose, and from other
> contributors. When I see conflicts in the code, I will state that clearly.


I interpreted you first mail:
http://www.xenomai.org/pipermail/xenomai/2012-July/000473.html
as the fact that you agreed to maintain the debian directory.

> 
>> Which is why, for the 2.6.2.1 release, I did not warn you (since it had
>> proved useless anyway)
> 
> This is also your interpretation of things, and your own decision. Not
> discussed with me.
> 
> Sorry when I missed some detail in the past. Please don't let this lead
> to bad further communication in return, or blaming each other.


I am not trying to blame you, I am trying to get a clear answer. If you
say you will maintain the debian directory, then I would not merge
patches touching it without your consent, if you say you do not, then I
can merge the changes.

> 
> Thanks,
> 
> Roland
> 
> 
> PS: I consider Debian itself the primary source of Debian packages and
> will continue to do so. When people ask specific questions about those,
> I'll answer accordingly.


This point of view is the point of view of the Debian community, but
when you answer on this mailing list, your answers can be interpreted as
coming from the "Xenomai community". And the Xenomai and Linux CNC
communities provide unofficial pre-built packages for several Debian
derivatives with the explicit purpose of avoiding newbies to have to ask
the same questions over and over again about compiling a kernel. So,
when you answer on this mailing list, it would be less disturbing if you
mentioned the solutions proposed by these communities. With all the
warnings you want about the packages not being official Debian packages.

> 
> PPS: If you consider your tone as helpful towards positive motivation of
> contributors, please rethink.

Everyone has its limitation, mine is to be unable to deal with nuances
in general and when communicating in particular. The only fix I know for
it is to avoid communicating. But when I read your answers to "Tom Z", I
can not really stay silent.

-- 
                                                                Gilles.

Re: [Xenomai] about Xenomai in Debian and Ubuntu

[Roland Stigge]

Hi Gilles,

On 03/13/2013 01:18 PM, Gilles Chanteperdrix wrote:
> Actually, what I quoted above is the contrary, you commanded me to warn
> you before a release, and I obeyed.
                            ^^^^^^^^
vs.

>>> Which is why, for the 2.6.2.1 release, I did not warn you (since it had
                                           ^^^^^^^^^^^^^^^^^^
>>> proved useless anyway)

Contradiction. _If_ you interpret my proposal _back_ _then_ as a command,
which I don't. ;-) "Command" is what I called you asking me for sth. and
blaming me of being inactive (for whatever reason - you can't tell if
someone you haven't met before in life is maybe struggling with sth.
else in life.)

> I interpreted you first mail:
> http://www.xenomai.org/pipermail/xenomai/2012-July/000473.html
> as the fact that you agreed to maintain the debian directory.

In this mail I actually stated that (1) it would be _possible_ for me to
do in Xenomai exactly _what I would change in Debian anyway_. and (2)
that I want to prevent duplicated work. So if I consider stuff in Debian
as working, I can't put much more effort into Xenomai's debian/, to do
different stuff you propose. Sorry - merely time restrictions - please
don't take it personal.

> I am not trying to blame you, I am trying to get a clear answer.
                                                     ^^^^^^^^^^^^
OK, under those circumstances of slightly diverging (i.e. minor but
conflicting in terms of project management) goals[1] I propose you commit
Xenomai's debian/* changes yourself without the need to wait for my comment.

In turn, I'm maintaining Debian's debian/* and synchronize as much as
possible (in the interest of both projects and their users). However,
there will be occasions where Xenomai's and Debian's debian/* will need
to diverge, which is fine, e.g. in case of Debian forces me to freeze
Debian, but apply security fixes etc., and Xenomai in turn wants to
support stuff that Debian can't always synchronize with.

Peace,

Roland


[1] Debian: freezing for releases, supporting Debian released versions
of Xenomai. Xenomai: Following up to date environments of different
Debian-like OSes, and commonly releasing more often.

Re: [Xenomai] about Xenomai in Debian and Ubuntu

[Leopold Palomo-Avellaneda]

Hi, (I think that I always put Hi or Bones (in catalan ..),

I would like to put some comments to this thread:

<Roland>
I really understand that you will take care of the debian directory of the 
upstream code. Probably I misunderstand something reading your posts ... I 
understand Gilles mail.

I propose to have two branches in the upstream tree, one with the necessary 
stuff to create a kernel package for the debian stable tree (with their 
kernel) and another for the sid/testing. This stuff could be the same of the 
official ones. Maybe some submodules or another hyper-special characteristic 
of git could help.

I don't understand your tone in your last mails. I think that it has been a 
misunderstanding problem. Please, _we_ can do it better.
</Roland>

<Gilles/Jan>
Yes, I understand to Stéphane. However as I'm in several projects, a lot of 
mails and to much things and a few time to do it. I have had the same feeling, 
but, well, maybe I begin to be a bit old and ....

I send to patch to UPSTREAM not to debian packagers. The script belongs to 
upstream. I had a bad feeling because no answer to the patch, no push. I had 
to insist because the stupid error. But, rereading the mail I think that I 
post it in a wrong and confusing way: mea culpa.
</Gilles/Jan>

Regards,

Leo
-- 
--
Linux User 152692
Catalonia

Re: [Xenomai] about Xenomai in Debian and Ubuntu

[Gilles Chanteperdrix]

On 03/13/2013 04:47 PM, Leopold Palomo-Avellaneda wrote:

> Hi, (I think that I always put Hi or Bones (in catalan ..),
> 
> I would like to put some comments to this thread:
> 
> <Roland>
> I really understand that you will take care of the debian directory of the 
> upstream code. Probably I misunderstand something reading your posts ... I 
> understand Gilles mail.
> 
> I propose to have two branches in the upstream tree, one with the necessary 
> stuff to create a kernel package for the debian stable tree (with their 
> kernel) and another for the sid/testing. This stuff could be the same of the 
> official ones. Maybe some submodules or another hyper-special characteristic 
> of git could help.
> 
> I don't understand your tone in your last mails. I think that it has been a 
> misunderstanding problem. Please, _we_ can do it better.
> </Roland>


> 
> <Gilles/Jan>
> Yes, I understand to Stéphane. However as I'm in several projects, a lot of 
> mails and to much things and a few time to do it. I have had the same feeling, 
> but, well, maybe I begin to be a bit old and ....
> 
> I send to patch to UPSTREAM not to debian packagers. The script belongs to 
> upstream. I had a bad feeling because no answer to the patch, no push. I had 
> to insist because the stupid error. But, rereading the mail I think that I 
> post it in a wrong and confusing way: mea culpa.
> </Gilles/Jan>


Hi Leopold,

Yes, there was a misunderstanding, but ultimately, I am the one who did
the release, so, I could have tried to fix things. The truth is that I
simply forgot the issue you reported. So, mea culpa.

As for maintaing two debian directories, this is a bad idea, as long as
we can make one which works for all releases, this is simply less work.
So, let us keep it that way.

Regards.

-- 
                                                                Gilles.

Re: [Xenomai] about Xenomai in Debian and Ubuntu

[Leopold Palomo-Avellaneda]

A Dimecres, 13 de març de 2013, Gilles Chanteperdrix va escriure:
> On 03/13/2013 04:47 PM, Leopold Palomo-Avellaneda wrote:
> 
> > Hi, (I think that I always put Hi or Bones (in catalan ..),
> > 
> > I would like to put some comments to this thread:
> > 
> > <Roland>
> > I really understand that you will take care of the debian directory of the 
> > upstream code. Probably I misunderstand something reading your posts ... I 
> > understand Gilles mail.
> > 
> > I propose to have two branches in the upstream tree, one with the 
necessary 
> > stuff to create a kernel package for the debian stable tree (with their 
> > kernel) and another for the sid/testing. This stuff could be the same of 
the 
> > official ones. Maybe some submodules or another hyper-special 
characteristic 
> > of git could help.
> > 
> > I don't understand your tone in your last mails. I think that it has been 
a 
> > misunderstanding problem. Please, _we_ can do it better.
> > </Roland>
> 
> 
> > 
> > <Gilles/Jan>
> > Yes, I understand to Stéphane. However as I'm in several projects, a lot 
of 
> > mails and to much things and a few time to do it. I have had the same 
feeling, 
> > but, well, maybe I begin to be a bit old and ....
> > 
> > I send to patch to UPSTREAM not to debian packagers. The script belongs to 
> > upstream. I had a bad feeling because no answer to the patch, no push. I 
had 
> > to insist because the stupid error. But, rereading the mail I think that I 
> > post it in a wrong and confusing way: mea culpa.
> > </Gilles/Jan>
> 
> 
> Hi Leopold,
> 
> Yes, there was a misunderstanding, but ultimately, I am the one who did
> the release, so, I could have tried to fix things. The truth is that I
> simply forgot the issue you reported. So, mea culpa.

too many guilty for a little thing ;-)

> As for maintaing two debian directories, this is a bad idea, as long as
> we can make one which works for all releases, this is simply less work.
> So, let us keep it that way.

Well, I explain myself. Normally _we_ have a stable version, with the needed 
kernel and sometimes, a more modern kernel, (backported) or in a 
testing/unstable version the modern kernel.

It's not easy sometimes to have a modern kernel in a stable version of debian. 
So, as now in the transitional time we should need two versions of the 
package. That's why my propose. 

OTOH, if we have _one_ sync with debian to me it's perfect. All of as we have 
a lot of things to do, so less work.

I hope that all could be solved.

Regards,

Leopold


-- 
--
Linux User 152692
Catalonia

Re: [Xenomai] about Xenomai in Debian and Ubuntu

[Gilles Chanteperdrix]

On 03/14/2013 01:16 PM, Leopold Palomo-Avellaneda wrote:

> A Dimecres, 13 de març de 2013, Gilles Chanteperdrix va escriure:
>> On 03/13/2013 04:47 PM, Leopold Palomo-Avellaneda wrote:
>>
>>> Hi, (I think that I always put Hi or Bones (in catalan ..),
>>>
>>> I would like to put some comments to this thread:
>>>
>>> <Roland>
>>> I really understand that you will take care of the debian directory of the 
>>> upstream code. Probably I misunderstand something reading your posts ... I 
>>> understand Gilles mail.
>>>
>>> I propose to have two branches in the upstream tree, one with the 
> necessary 
>>> stuff to create a kernel package for the debian stable tree (with their 
>>> kernel) and another for the sid/testing. This stuff could be the same of 
> the 
>>> official ones. Maybe some submodules or another hyper-special 
> characteristic 
>>> of git could help.
>>>
>>> I don't understand your tone in your last mails. I think that it has been 
> a 
>>> misunderstanding problem. Please, _we_ can do it better.
>>> </Roland>
>>
>>
>>>
>>> <Gilles/Jan>
>>> Yes, I understand to Stéphane. However as I'm in several projects, a lot 
> of 
>>> mails and to much things and a few time to do it. I have had the same 
> feeling, 
>>> but, well, maybe I begin to be a bit old and ....
>>>
>>> I send to patch to UPSTREAM not to debian packagers. The script belongs to 
>>> upstream. I had a bad feeling because no answer to the patch, no push. I 
> had 
>>> to insist because the stupid error. But, rereading the mail I think that I 
>>> post it in a wrong and confusing way: mea culpa.
>>> </Gilles/Jan>
>>
>>
>> Hi Leopold,
>>
>> Yes, there was a misunderstanding, but ultimately, I am the one who did
>> the release, so, I could have tried to fix things. The truth is that I
>> simply forgot the issue you reported. So, mea culpa.
> 
> too many guilty for a little thing ;-)
> 
>> As for maintaing two debian directories, this is a bad idea, as long as
>> we can make one which works for all releases, this is simply less work.
>> So, let us keep it that way.
> 
> Well, I explain myself. Normally _we_ have a stable version, with the needed 
> kernel and sometimes, a more modern kernel, (backported) or in a 
> testing/unstable version the modern kernel.
> 
> It's not easy sometimes to have a modern kernel in a stable version of debian. 
> So, as now in the transitional time we should need two versions of the 
> package. That's why my propose. 
> 
> OTOH, if we have _one_ sync with debian to me it's perfect. All of as we have 
> a lot of things to do, so less work.
> 
> I hope that all could be solved.


Hi Leopold,

quite frankly I do not understand what you are talking about. What I am
talking about is the "debian" directory in xenomai sources. It seems
simpler to me to have only one directory which works for all Debian
branches, in order to allow users to generate packages for the
distribution they want. That is all I am saying.

Regards.

-- 
                                                                Gilles.

Re: [Xenomai] about Xenomai in Debian and Ubuntu

[Leopold Palomo-Avellaneda]

A Divendres, 15 de març de 2013, Gilles Chanteperdrix va escriure:
> On 03/14/2013 01:16 PM, Leopold Palomo-Avellaneda wrote:
> 
> > A Dimecres, 13 de març de 2013, Gilles Chanteperdrix va escriure:
> >> On 03/13/2013 04:47 PM, Leopold Palomo-Avellaneda wrote:
> >>
> >>> Hi, (I think that I always put Hi or Bones (in catalan ..),
> >>>
> >>> I would like to put some comments to this thread:
> >>>
> >>> <Roland>
> >>> I really understand that you will take care of the debian directory of 
the 
> >>> upstream code. Probably I misunderstand something reading your posts ... 
I 
> >>> understand Gilles mail.
> >>>
> >>> I propose to have two branches in the upstream tree, one with the 
> > necessary 
> >>> stuff to create a kernel package for the debian stable tree (with their 
> >>> kernel) and another for the sid/testing. This stuff could be the same of 
> > the 
> >>> official ones. Maybe some submodules or another hyper-special 
> > characteristic 
> >>> of git could help.
> >>>
> >>> I don't understand your tone in your last mails. I think that it has 
been 
> > a 
> >>> misunderstanding problem. Please, _we_ can do it better.
> >>> </Roland>
> >>
> >>
> >>>
> >>> <Gilles/Jan>
> >>> Yes, I understand to Stéphane. However as I'm in several projects, a lot 
> > of 
> >>> mails and to much things and a few time to do it. I have had the same 
> > feeling, 
> >>> but, well, maybe I begin to be a bit old and ....
> >>>
> >>> I send to patch to UPSTREAM not to debian packagers. The script belongs 
to 
> >>> upstream. I had a bad feeling because no answer to the patch, no push. I 
> > had 
> >>> to insist because the stupid error. But, rereading the mail I think that 
I 
> >>> post it in a wrong and confusing way: mea culpa.
> >>> </Gilles/Jan>
> >>
> >>
> >> Hi Leopold,
> >>
> >> Yes, there was a misunderstanding, but ultimately, I am the one who did
> >> the release, so, I could have tried to fix things. The truth is that I
> >> simply forgot the issue you reported. So, mea culpa.
> > 
> > too many guilty for a little thing ;-)
> > 
> >> As for maintaing two debian directories, this is a bad idea, as long as
> >> we can make one which works for all releases, this is simply less work.
> >> So, let us keep it that way.
> > 
> > Well, I explain myself. Normally _we_ have a stable version, with the 
needed 
> > kernel and sometimes, a more modern kernel, (backported) or in a 
> > testing/unstable version the modern kernel.
> > 
> > It's not easy sometimes to have a modern kernel in a stable version of 
debian. 
> > So, as now in the transitional time we should need two versions of the 
> > package. That's why my propose. 
> > 
> > OTOH, if we have _one_ sync with debian to me it's perfect. All of as we 
have 
> > a lot of things to do, so less work.
> > 
> > I hope that all could be solved.
> 
> 
> Hi Leopold,
> 
> quite frankly I do not understand what you are talking about. 

:-)

> What I am
> talking about is the "debian" directory in xenomai sources. 

yes, I was talking about the same.

> It seems
> simpler to me to have only one directory which works for all Debian
> branches, in order to allow users to generate packages for the
> distribution they want. That is all I am saying.

if it works, to me it's ok. But I don't know if you can have _one_ directory 
for all Debian branches. For example in the control file:

Build-Depends: debhelper (>= 8), findutils (>= 4.2.28), autotools-dev, 
autoconf, automake, libtool

I cannot predict if jezzie will introduce a change in debhelper that breaks 
wheezy. Or findutils could have a very different version in wheezy than 
squeeze. Or simple, we could have different versions (backports, testing, 
experimental) for the same sources and the changelog file will be a mesh. 

It's just and idea. But, taking in mind how many users uses it and the amount 
of work that represents it, please do it the most simplest way.

Thanks for all.

Regards.


Leopold 


> 
> -- 
>                                                                 Gilles.
> 
> 


-- 
--
Linux User 152692
Catalonia

Re: [Xenomai] about Xenomai in Debian and Ubuntu

[Gilles Chanteperdrix]

On 03/19/2013 09:33 AM, Leopold Palomo-Avellaneda wrote:

> A Divendres, 15 de març de 2013, Gilles Chanteperdrix va escriure:
>> On 03/14/2013 01:16 PM, Leopold Palomo-Avellaneda wrote:
>>
>>> A Dimecres, 13 de març de 2013, Gilles Chanteperdrix va escriure:
>>>> On 03/13/2013 04:47 PM, Leopold Palomo-Avellaneda wrote:
>>>>
>>>>> Hi, (I think that I always put Hi or Bones (in catalan ..),
>>>>>
>>>>> I would like to put some comments to this thread:
>>>>>
>>>>> <Roland>
>>>>> I really understand that you will take care of the debian directory of 
> the 
>>>>> upstream code. Probably I misunderstand something reading your posts ... 
> I 
>>>>> understand Gilles mail.
>>>>>
>>>>> I propose to have two branches in the upstream tree, one with the 
>>> necessary 
>>>>> stuff to create a kernel package for the debian stable tree (with their 
>>>>> kernel) and another for the sid/testing. This stuff could be the same of 
>>> the 
>>>>> official ones. Maybe some submodules or another hyper-special 
>>> characteristic 
>>>>> of git could help.
>>>>>
>>>>> I don't understand your tone in your last mails. I think that it has 
> been 
>>> a 
>>>>> misunderstanding problem. Please, _we_ can do it better.
>>>>> </Roland>
>>>>
>>>>
>>>>>
>>>>> <Gilles/Jan>
>>>>> Yes, I understand to Stéphane. However as I'm in several projects, a lot 
>>> of 
>>>>> mails and to much things and a few time to do it. I have had the same 
>>> feeling, 
>>>>> but, well, maybe I begin to be a bit old and ....
>>>>>
>>>>> I send to patch to UPSTREAM not to debian packagers. The script belongs 
> to 
>>>>> upstream. I had a bad feeling because no answer to the patch, no push. I 
>>> had 
>>>>> to insist because the stupid error. But, rereading the mail I think that 
> I 
>>>>> post it in a wrong and confusing way: mea culpa.
>>>>> </Gilles/Jan>
>>>>
>>>>
>>>> Hi Leopold,
>>>>
>>>> Yes, there was a misunderstanding, but ultimately, I am the one who did
>>>> the release, so, I could have tried to fix things. The truth is that I
>>>> simply forgot the issue you reported. So, mea culpa.
>>>
>>> too many guilty for a little thing ;-)
>>>
>>>> As for maintaing two debian directories, this is a bad idea, as long as
>>>> we can make one which works for all releases, this is simply less work.
>>>> So, let us keep it that way.
>>>
>>> Well, I explain myself. Normally _we_ have a stable version, with the 
> needed 
>>> kernel and sometimes, a more modern kernel, (backported) or in a 
>>> testing/unstable version the modern kernel.
>>>
>>> It's not easy sometimes to have a modern kernel in a stable version of 
> debian. 
>>> So, as now in the transitional time we should need two versions of the 
>>> package. That's why my propose. 
>>>
>>> OTOH, if we have _one_ sync with debian to me it's perfect. All of as we 
> have 
>>> a lot of things to do, so less work.
>>>
>>> I hope that all could be solved.
>>
>>
>> Hi Leopold,
>>
>> quite frankly I do not understand what you are talking about. 
> 
> :-)
> 
>> What I am
>> talking about is the "debian" directory in xenomai sources. 
> 
> yes, I was talking about the same.
> 
>> It seems
>> simpler to me to have only one directory which works for all Debian
>> branches, in order to allow users to generate packages for the
>> distribution they want. That is all I am saying.
> 
> if it works, to me it's ok. But I don't know if you can have _one_ directory 
> for all Debian branches. For example in the control file:
> 
> Build-Depends: debhelper (>= 8), findutils (>= 4.2.28), autotools-dev, 
> autoconf, automake, libtool
> 
> I cannot predict if jezzie will introduce a change in debhelper that breaks 
> wheezy. Or findutils could have a very different version in wheezy than 
> squeeze. Or simple, we could have different versions (backports, testing, 
> experimental) for the same sources and the changelog file will be a mesh. 
> 
> It's just and idea. But, taking in mind how many users uses it and the amount 
> of work that represents it, please do it the most simplest way.


I do not think depending on too old versions really matters. I happen to
regularly compile on squeeze packages not for squeeze, and most of the
time, simply modifying the control file to require lower versions simply
works. So, conversely having a control file requesting too old versions
should not be a problem.

I agree that there may be other reasons why the package could break on
more recent versions, if that happens and someone actually has the
problem, we will get a bug report and react accordingly.

Note that the debian directory is mostly for the stable branch, as
Roland takes care of debian packages for the other branches.

-- 
                                                                Gilles.

Re: [Xenomai] about Xenomai in Debian and Ubuntu

[Gilles Chanteperdrix]

On 03/13/2013 01:53 PM, Roland Stigge wrote:
> Hi Gilles,

Hi Roland,

> 
>> I am not trying to blame you, I am trying to get a clear answer.
>                                                      ^^^^^^^^^^^^
> OK, under those circumstances of slightly diverging (i.e. minor but
> conflicting in terms of project management) goals[1] I propose you commit
> Xenomai's debian/* changes yourself without the need to wait for my comment.


Ack.

As a side note, each xenomai branch really is a "stable" branch, any
kernel/user combinations of any two releases in the same branch are API
and binary compatible, so, it would seem possible to consider every
release as a security fix for the previous.

Regards.

-- 
                                                                Gilles.