Handling of system bundles with shell commands

[cschneider]

Hi all,

currently when someone accesses a system bundle using a shell command we
ask if he is sure and he has to type "yes" or "no". We also have the
force option so yes is assumed.

I propose we change the behaviour for karaf 3.0:

If force is not given then the command should simply fail with an error
if system bundles are accessed. If force is set it should work on the
system bundles. So we have the same
safety for handling system bundles and the user can simply give -f on
the second try.

Christian

--
Christian Schneider
http://www.liquid-reality.de

Open Source Architect
Talend Application Integration Division http://www.talend.com

[Andreas Pieber]

+1 if we could change it quickly; otherwise +1 but delay it for some
later release. IMHO to get 3.x out has a higher priority than if you
use yes/no or -f

Kind regards,
Andreas

On Sun, Mar 18, 2012 at 12:38, Christian Schneider
<[hidden email]> wrote:

> Hi all,
>
> currently when someone accesses a system bundle using a shell command we ask
> if he is sure and he has to type "yes" or "no". We also have the force
> option so yes is assumed.
>
> I propose we change the behaviour for karaf 3.0:
>
> If force is not given then the command should simply fail with an error if
> system bundles are accessed. If force is set it should work on the system
> bundles. So we have the same
> safety for handling system bundles and the user can simply give -f on the
> second try.
>
> Christian
>
> --
> Christian Schneider
> http://www.liquid-reality.de
>
> Open Source Architect
> Talend Application Integration Division http://www.talend.com
>

[jbonofre]

+1 as far as we provide a clean and understandable message to the users.

Regards
JB

On 03/18/2012 12:38 PM, Christian Schneider wrote:

> Hi all,
>
> currently when someone accesses a system bundle using a shell command we
> ask if he is sure and he has to type "yes" or "no". We also have the
> force option so yes is assumed.
>
> I propose we change the behaviour for karaf 3.0:
>
> If force is not given then the command should simply fail with an error
> if system bundles are accessed. If force is set it should work on the
> system bundles. So we have the same
> safety for handling system bundles and the user can simply give -f on
> the second try.
>
> Christian
>

--
Jean-Baptiste Onofré
[hidden email]
http://blog.nanthrax.net
Talend - http://www.talend.com

[Achim Nierbeck]

+/-0
don't know if we really need to change it, it doesn't give me that much
pain right now, even when testing
pax-web bundles and these are a lot ;)
if we decide that it gives us enough pain for a -f than I'm with Andreas
that it's either a quick one or should be postponed :)

regards, Achim

Am 18.03.2012 15:29, schrieb Andreas Pieber:

> +1 if we could change it quickly; otherwise +1 but delay it for some
> later release. IMHO to get 3.x out has a higher priority than if you
> use yes/no or -f
>
> Kind regards,
> Andreas
>
> On Sun, Mar 18, 2012 at 12:38, Christian Schneider
> <[hidden email]>  wrote:
>> Hi all,
>>
>> currently when someone accesses a system bundle using a shell command we ask
>> if he is sure and he has to type "yes" or "no". We also have the force
>> option so yes is assumed.
>>
>> I propose we change the behaviour for karaf 3.0:
>>
>> If force is not given then the command should simply fail with an error if
>> system bundles are accessed. If force is set it should work on the system
>> bundles. So we have the same
>> safety for handling system bundles and the user can simply give -f on the
>> second try.
>>
>> Christian
>>
>> --
>> Christian Schneider
>> http://www.liquid-reality.de
>>
>> Open Source Architect
>> Talend Application Integration Division http://www.talend.com
>>

--
- Apache Karaf<http://karaf.apache.org/>  Committer&  PMC
- OPS4J Pax Web<http://wiki.ops4j.org/display/paxweb/Pax+Web/>    Committer&  Project Lead
- Blog<http://notizblog.nierbeck.de/>

[jbonofre]

Agree with both Andreas and Achim.

We didn't get any comments from users around that, so I guess that we
don't have any emergency.

Regards
JB

On 03/18/2012 06:48 PM, Achim Nierbeck wrote:

> +/-0
> don't know if we really need to change it, it doesn't give me that much
> pain right now, even when testing
> pax-web bundles and these are a lot ;)
> if we decide that it gives us enough pain for a -f than I'm with Andreas
> that it's either a quick one or should be postponed :)
>
> regards, Achim
>
> Am 18.03.2012 15:29, schrieb Andreas Pieber:
>> +1 if we could change it quickly; otherwise +1 but delay it for some
>> later release. IMHO to get 3.x out has a higher priority than if you
>> use yes/no or -f
>>
>> Kind regards,
>> Andreas
>>
>> On Sun, Mar 18, 2012 at 12:38, Christian Schneider
>> <[hidden email]> wrote:
>>> Hi all,
>>>
>>> currently when someone accesses a system bundle using a shell command
>>> we ask
>>> if he is sure and he has to type "yes" or "no". We also have the force
>>> option so yes is assumed.
>>>
>>> I propose we change the behaviour for karaf 3.0:
>>>
>>> If force is not given then the command should simply fail with an
>>> error if
>>> system bundles are accessed. If force is set it should work on the
>>> system
>>> bundles. So we have the same
>>> safety for handling system bundles and the user can simply give -f on
>>> the
>>> second try.
>>>
>>> Christian
>>>
>>> --
>>> Christian Schneider
>>> http://www.liquid-reality.de
>>>
>>> Open Source Architect
>>> Talend Application Integration Division http://www.talend.com
>>>
>
>

--
Jean-Baptiste Onofré
[hidden email]
http://blog.nanthrax.net
Talend - http://www.talend.com

[cschneider]

It is an easy fix. So I am sure I can make it on time.

One advantage is that we have a clean spearation of core and command. I
am currently moving the bundle selector logic to a core module for
bundle. Having a reference from a core class to Session is not so good.
I had already prepared to split the filtering for system bundles out of
the selector and move it to the command layer as it needs user
interaction. Much better would be to not need interaction though.

Christian


Am 18.03.2012 18:51, schrieb Jean-Baptiste Onofré:

> Agree with both Andreas and Achim.
>
> We didn't get any comments from users around that, so I guess that we
> don't have any emergency.
>
> Regards
> JB
>
> On 03/18/2012 06:48 PM, Achim Nierbeck wrote:
>> +/-0
>> don't know if we really need to change it, it doesn't give me that much
>> pain right now, even when testing
>> pax-web bundles and these are a lot ;)
>> if we decide that it gives us enough pain for a -f than I'm with Andreas
>> that it's either a quick one or should be postponed :)
>>
>> regards, Achim
>>
>> Am 18.03.2012 15:29, schrieb Andreas Pieber:
>>> +1 if we could change it quickly; otherwise +1 but delay it for some
>>> later release. IMHO to get 3.x out has a higher priority than if you
>>> use yes/no or -f
>>>
>>> Kind regards,
>>> Andreas
>>>
>>> On Sun, Mar 18, 2012 at 12:38, Christian Schneider
>>> <[hidden email]> wrote:
>>>> Hi all,
>>>>
>>>> currently when someone accesses a system bundle using a shell command
>>>> we ask
>>>> if he is sure and he has to type "yes" or "no". We also have the force
>>>> option so yes is assumed.
>>>>
>>>> I propose we change the behaviour for karaf 3.0:
>>>>
>>>> If force is not given then the command should simply fail with an
>>>> error if
>>>> system bundles are accessed. If force is set it should work on the
>>>> system
>>>> bundles. So we have the same
>>>> safety for handling system bundles and the user can simply give -f on
>>>> the
>>>> second try.
>>>>
>>>> Christian
>>>>
>>>> --
>>>> Christian Schneider
>>>> http://www.liquid-reality.de
>>>>
>>>> Open Source Architect
>>>> Talend Application Integration Division http://www.talend.com
>>>>
>>
>>
>

--
Christian Schneider
http://www.liquid-reality.de

Open Source Architect
Talend Application Integration Division http://www.talend.com

[jbonofre]

It sounds good, but I will do that for Karaf 3.0.0 (so on trunk), not on
Karaf 2.2.x.

Regards
JB

On 03/18/2012 07:57 PM, Christian Schneider wrote:

> It is an easy fix. So I am sure I can make it on time.
>
> One advantage is that we have a clean spearation of core and command. I
> am currently moving the bundle selector logic to a core module for
> bundle. Having a reference from a core class to Session is not so good.
> I had already prepared to split the filtering for system bundles out of
> the selector and move it to the command layer as it needs user
> interaction. Much better would be to not need interaction though.
>
> Christian
>
>
> Am 18.03.2012 18:51, schrieb Jean-Baptiste Onofré:
>> Agree with both Andreas and Achim.
>>
>> We didn't get any comments from users around that, so I guess that we
>> don't have any emergency.
>>
>> Regards
>> JB
>>
>> On 03/18/2012 06:48 PM, Achim Nierbeck wrote:
>>> +/-0
>>> don't know if we really need to change it, it doesn't give me that much
>>> pain right now, even when testing
>>> pax-web bundles and these are a lot ;)
>>> if we decide that it gives us enough pain for a -f than I'm with Andreas
>>> that it's either a quick one or should be postponed :)
>>>
>>> regards, Achim
>>>
>>> Am 18.03.2012 15:29, schrieb Andreas Pieber:
>>>> +1 if we could change it quickly; otherwise +1 but delay it for some
>>>> later release. IMHO to get 3.x out has a higher priority than if you
>>>> use yes/no or -f
>>>>
>>>> Kind regards,
>>>> Andreas
>>>>
>>>> On Sun, Mar 18, 2012 at 12:38, Christian Schneider
>>>> <[hidden email]> wrote:
>>>>> Hi all,
>>>>>
>>>>> currently when someone accesses a system bundle using a shell command
>>>>> we ask
>>>>> if he is sure and he has to type "yes" or "no". We also have the force
>>>>> option so yes is assumed.
>>>>>
>>>>> I propose we change the behaviour for karaf 3.0:
>>>>>
>>>>> If force is not given then the command should simply fail with an
>>>>> error if
>>>>> system bundles are accessed. If force is set it should work on the
>>>>> system
>>>>> bundles. So we have the same
>>>>> safety for handling system bundles and the user can simply give -f on
>>>>> the
>>>>> second try.
>>>>>
>>>>> Christian
>>>>>
>>>>> --
>>>>> Christian Schneider
>>>>> http://www.liquid-reality.de
>>>>>
>>>>> Open Source Architect
>>>>> Talend Application Integration Division http://www.talend.com
>>>>>
>>>
>>>
>>
>
>

--
Jean-Baptiste Onofré
[hidden email]
http://blog.nanthrax.net
Talend - http://www.talend.com

[cschneider]

I have committed the change for karaf 3. I do not plan to backpot to 2.2.x.

The behaviour is now lik this:
karaf@root> bundle:stop 1
Error executing command: Access to system bundle 1 denied. You can
override with -f

Christian

Am 18.03.2012 21:05, schrieb Jean-Baptiste Onofré:

> It sounds good, but I will do that for Karaf 3.0.0 (so on trunk), not
> on Karaf 2.2.x.
>
> Regards
> JB
>
> On 03/18/2012 07:57 PM, Christian Schneider wrote:
>> It is an easy fix. So I am sure I can make it on time.
>>
>> One advantage is that we have a clean spearation of core and command. I
>> am currently moving the bundle selector logic to a core module for
>> bundle. Having a reference from a core class to Session is not so good.
>> I had already prepared to split the filtering for system bundles out of
>> the selector and move it to the command layer as it needs user
>> interaction. Much better would be to not need interaction though.
>>
>> Christian
>>
>>
>> Am 18.03.2012 18:51, schrieb Jean-Baptiste Onofré:
>>> Agree with both Andreas and Achim.
>>>
>>> We didn't get any comments from users around that, so I guess that we
>>> don't have any emergency.
>>>
>>> Regards
>>> JB
>>>
>>> On 03/18/2012 06:48 PM, Achim Nierbeck wrote:
>>>> +/-0
>>>> don't know if we really need to change it, it doesn't give me that
>>>> much
>>>> pain right now, even when testing
>>>> pax-web bundles and these are a lot ;)
>>>> if we decide that it gives us enough pain for a -f than I'm with
>>>> Andreas
>>>> that it's either a quick one or should be postponed :)
>>>>
>>>> regards, Achim
>>>>
>>>> Am 18.03.2012 15:29, schrieb Andreas Pieber:
>>>>> +1 if we could change it quickly; otherwise +1 but delay it for some
>>>>> later release. IMHO to get 3.x out has a higher priority than if you
>>>>> use yes/no or -f
>>>>>
>>>>> Kind regards,
>>>>> Andreas
>>>>>
>>>>> On Sun, Mar 18, 2012 at 12:38, Christian Schneider
>>>>> <[hidden email]> wrote:
>>>>>> Hi all,
>>>>>>
>>>>>> currently when someone accesses a system bundle using a shell
>>>>>> command
>>>>>> we ask
>>>>>> if he is sure and he has to type "yes" or "no". We also have the
>>>>>> force
>>>>>> option so yes is assumed.
>>>>>>
>>>>>> I propose we change the behaviour for karaf 3.0:
>>>>>>
>>>>>> If force is not given then the command should simply fail with an
>>>>>> error if
>>>>>> system bundles are accessed. If force is set it should work on the
>>>>>> system
>>>>>> bundles. So we have the same
>>>>>> safety for handling system bundles and the user can simply give
>>>>>> -f on
>>>>>> the
>>>>>> second try.
>>>>>>
>>>>>> Christian
>>>>>>
>>>>>> --
>>>>>> Christian Schneider
>>>>>> http://www.liquid-reality.de
>>>>>>
>>>>>> Open Source Architect
>>>>>> Talend Application Integration Division http://www.talend.com
>>>>>>
>>>>
>>>>
>>>
>>
>>
>

--
Christian Schneider
http://www.liquid-reality.de

Open Source Architect
Talend Application Integration Division http://www.talend.com

[jbonofre]

Hi Christian,

it sounds good to me ;)

Thanks,
Regards
JB

On Wed 21/03/12 10:59 , Christian Schneider  wrote::

I have committed the change for karaf 3. I do not plan to backpot to 2.2.x.

The behaviour is now lik this:
karaf@root> bundle:stop 1
Error executing command: Access to system bundle 1 denied. You can
override with -f

Christian

Am 18.03.2012 21:05, schrieb Jean-Baptiste Onofré:

> It sounds good, but I will do that for Karaf 3.0.0 (so on trunk), not
> on Karaf 2.2.x.
>
> Regards
> JB
>
> On 03/18/2012 07:57 PM, Christian Schneider wrote:
>> It is an easy fix. So I am sure I can make it on time.
>>
>> One advantage is that we have a clean spearation of core and command. I
>> am currently moving the bundle selector logic to a core module for
>> bundle. Having a reference from a core class to Session is not so good.
>> I had already prepared to split the filtering for system bundles out of
>> the selector and move it to the command layer as it needs user
>> interaction. Much better would be to not need interaction though.
>>
>> Christian
>>
>>
>> Am 18.03.2012 18:51, schrieb Jean-Baptiste Onofré:
>>> Agree with both Andreas and Achim.
>>>
>>> We didn't get any comments from users around that, so I guess that we
>>> don't have any emergency.
>>>
>>> Regards
>>> JB
>>>
>>> On 03/18/2012 06:48 PM, Achim Nierbeck wrote:
>>>> +/-0
>>>> don't know if we really need to change it, it doesn't give me that
>>>> much
>>>> pain right now, even when testing
>>>> pax-web bundles and these are a lot ;)
>>>> if we decide that it gives us enough pain for a -f than I'm with
>>>> Andreas
>>>> that it's either a quick one or should be postponed :)
>>>>
>>>> regards, Achim
>>>>
>>>> Am 18.03.2012 15:29, schrieb Andreas Pieber:
>>>>> +1 if we could change it quickly; otherwise +1 but delay it for some
>>>>> later release. IMHO to get 3.x out has a higher priority than if you
>>>>> use yes/no or -f
>>>>>
>>>>> Kind regards,
>>>>> Andreas
>>>>>
>>>>> On Sun, Mar 18, 2012 at 12:38, Christian Schneider
>>>>> [hidden email]> wrote:
>>>>>> Hi all,
>>>>>>
>>>>>> currently when someone accesses a system bundle using a shell
>>>>>> command
>>>>>> we ask
>>>>>> if he is sure and he has to type "yes" or "no". We also have the
>>>>>> force
>>>>>> option so yes is assumed.
>>>>>>
>>>>>> I propose we change the behaviour for karaf 3.0:
>>>>>>
>>>>>> If force is not given then the command should simply fail with an
>>>>>> error if
>>>>>> system bundles are accessed. If force is set it should work on the
>>>>>> system
>>>>>> bundles. So we have the same
>>>>>> safety for handling system bundles and the user can simply give
>>>>>> -f on
>>>>>> the
>>>>>> second try.
>>>>>>
>>>>>> Christian
>>>>>>
>>>>>> --
>>>>>> Christian Schneider
>>>>>> http://www.liquid-reality.de>
>>>>>>
>>>>>> Open Source Architect
>>>>>> Talend Application Integration Division http://www.talend.com>
>>>>>>
>>>>
>>>>
>>>
>>
>>
>

--
Christian Schneider
http://www.liquid-reality.de

Open Source Architect
Talend Application Integration Division http://www.talend.com