[Rock-dev] The issue with Rock
Sylvain Joyeux
bir.sylvain at gmail.com
Wed Jun 18 15:16:08 CEST 2014
One possible solution: decouple rock master from rock.core master
(different release schedules). rock.core master would then become really a
development branch where people should not fear to push tested but not
heavily tested code.
The rules:
1. rock master CANNOT depend on features in rock.core master. This ensures
that the current up-to-date packages are not dependent on development
features in the toolchain. This is ensured by making people always use
rock.core next or stable.
2. rock.core master MUST stay backward compatible, i.e. at the point of
release of rock.core, rock next should be usable with rock.core master. In
other words, releasing rock.core (from master to next) should not break
existing 'next' applications.
Thoughts?
Sylvain
On Thu, Jun 5, 2014 at 2:02 PM, Matthias Goldhoorn <
matthias.goldhoorn at dfki.de> wrote:
> -1 because for checkint our tree we need master, on we simply remove it
> from the server i'm sure people find ways like in the overrides
>
> *:
> - branch: master
>
> to do such things.
>
> Best,
> Matthias
>
>
> On 05.06.2014 13:23, Vincent vittori wrote:
>
> +1 for removing master
>
>
> 2014-06-05 9:25 GMT+02:00 Alexander Duda <Alexander.Duda at dfki.de>:
>
>>
>> Am 05.06.2014 um 09:08 schrieb Matthias Goldhoorn <
>> matthias.goldhoorn at dfki.de>:
>>
>> Good Morning,
>>
>> I don't know what you mean with resistance, in the past master was the
>> most "stable" version with the newest features.
>> Therefore there was no resistance.
>>
>> Specially for Syskit/Roby i not posted all bugs, most of the bugs i
>> workaround, but i will start to create bug-requests for this...
>>
>> Regarding the new functionality, you are right, most of these features
>> are not needed, they make only the life easier..., but if i as developer
>> see the need of a feature and implement it, i want to use this directly. I
>> think this is normal, since the rock-devs are not pure-rock devs, work is
>> done if we feel that we need enhancements...
>>
>> Maybe we should rename the structure or introduce a experimental branch.
>> Phsychological i have the impression that "master" gets associated with
>> "newest" not with an "unstable development" version. Maybe we should rename
>> or create an additional "unstable" branch, from where the release policy to
>> master is not so fixed windowed...
>>
>> example development:
>> - work is done on experimental and pushed as soon as the dev thing it
>> might work
>> - experimental is pushed to master, as soon the responsible dev thing his
>> changes work for all (few days upto a week?)
>>
>> Again i thing the primary reasons why most of all stays on master ist
>> that the other branches does not have the "needed"("wished") features, or
>> they have bugs that are not pushed from master
>>
>>
>> I have my doubt here. None of the core packages had major issues the
>> last couple of months forcing people to use master for the hole bootstrap.
>>
>>
>> Thoughts?
>>
>>
>> I have the feeling we should just remove master as flavor. In this case
>> everyone has to bootstrap next or stable and if needed he/she can manually
>> overwrite specific packages.
>>
>> Alex
>>
>>
>> Matthias
>>
>>
>>
>> On 04.06.2014 17:00, Sylvain Joyeux wrote:
>>
>> Then ... my next question would be:
>>
>> Why isn't there more resistance w.r.t. switching to master ?
>>
>> I mean, when you say "oh I had a bug on syskit on next", did you report
>> it as a functionality bug on next ? Did you insist that it should be fixed
>> *on next* ? instead of switching to master ?
>>
>> For new functionality, how much of it is "oh but I need X, it is so
>> shiny" instead of "without X, I really cannot do it !". I mean, when I
>> worked on the Orion I *wanted* some features from master, but quickly
>> realized that I did not *need* them. I had what was strictly needed to get
>> the Joints type (meaning typelib/master but orogen/next)
>>
>> As for the release schedule / frequency, I can only do +1. Releases are
>> too far apart.
>>
>> My big problem here is that master has become the de-facto version of
>> Rock that everyone uses, which really hinders possibility to do some actual
>> development.
>>
>> Sylvain
>>
>>
>> On Wed, Jun 4, 2014 at 3:54 PM, Matthias Goldhoorn <
>> matthias.goldhoorn at dfki.de> wrote:
>>
>>> On 04.06.2014 15:43, Sylvain Joyeux wrote:
>>>
>>> Is that everyone seem to think that they need master. The majority
>>> should be using stable or next.
>>>
>>> Now, I *know* that there are reasons (there are always reasons) why
>>> one might think that master is required. However, the main question for me
>>> is:
>>>
>>> How can we make people feel confident that they can use next ?
>>>
>>> Or
>>>
>>> How can we ensure that 'next' can be used except for a few packages
>>> that would go on master ?
>>>
>>> The best way to start answering these questions is to answer another
>>> one:
>>>
>>> Why are you on master ?
>>>
>>>
>>> Because i using syskit and the next version is even more unstable than
>>> master. I had several times that the depandancy between roby/syskit and
>>> other 'core' packages is hard. So i cannot stay a long time on next and
>>> only with syskit/roby on master.
>>> Indeed i'm not sure if i can currently use syskit/roby on next and
>>> everything else on master.
>>>
>>> So generally speaking, incompatibilities between
>>> syskit/roby/utilmm/utilrb/typelib/orogen/ base/types/(std)
>>>
>>>
>>> The Second point, is that the release cycle to next is to long for new
>>> features, i i (as rock-dev) add new features to rock. I take ofter months
>>> before it goes into next.
>>> Therefore i have (due to the same reasons above) switch to master, also
>>> for other members of my project. I would prefer a shorter release time
>>> between master/stable/next...
>>>
>>>
>>>
>>> Best,
>>> Matthias
>>>
>>>
>>>
>>>
>>> Sylvain
>>>
>>>
>>> _______________________________________________
>>> Rock-dev mailing listRock-dev at dfki.dehttp://www.dfki.de/mailman/cgi-bin/listinfo/rock-dev
>>>
>>>
>>>
>>> --
>>> --
>>> Matthias Goldhoorn
>>> Unterwasserrobotik
>>>
>>> Standort Bremen:
>>> DFKI GmbH
>>> Robotics Innovation Center
>>> Robert-Hooke-Straße 5
>>> 28359 Bremen, Germany
>>>
>>> Phone: +49 (0)421 218-64100
>>> Fax: +49 (0)421 218-64150
>>> E-Mail: robotik at dfki.de
>>>
>>> Weitere Informationen: http://www.dfki.de/robotik
>>> -----------------------------------------------------------------------
>>> Deutsches Forschungszentrum fuer Kuenstliche Intelligenz GmbH
>>> Firmensitz: Trippstadter Straße 122, D-67663 Kaiserslautern
>>> Geschaeftsfuehrung: Prof. Dr. Dr. h.c. mult. Wolfgang Wahlster
>>> (Vorsitzender) Dr. Walter Olthoff
>>> Vorsitzender des Aufsichtsrats: Prof. Dr. h.c. Hans A. Aukes
>>> Amtsgericht Kaiserslautern, HRB 2313
>>> Sitz der Gesellschaft: Kaiserslautern (HRB 2313)
>>> USt-Id.Nr.: DE 148646973
>>> Steuernummer: 19/673/0060/3
>>> -----------------------------------------------------------------------
>>>
>>>
>>>
>>>
>>
>>
>> --
>> Dipl.-Inf. Matthias Goldhoorn
>> Space and Underwater Robotic
>>
>> Universität Bremen
>> FB 3 - Mathematik und Informatik
>> AG Robotik
>> Robert-Hooke-Straße 1
>> 28359 Bremen, Germany
>>
>> Zentrale: +49 421 178 45-6611
>>
>> Besuchsadresse der Nebengeschäftstelle:
>> Robert-Hooke-Straße 5
>> 28359 Bremen, Germany
>>
>> Tel.: +49 421 178 45-4193
>> Empfang: +49 421 178 45-6600
>> Fax: +49 421 178 45-4150
>> E-Mail: matthias.goldhoorn at informatik.uni-bremen.de
>>
>> Weitere Informationen: http://www.informatik.uni-bremen.de/robotik
>>
>> _______________________________________________
>> Rock-dev mailing list
>> Rock-dev at dfki.de
>> http://www.dfki.de/mailman/cgi-bin/listinfo/rock-dev
>>
>>
>>
>> --
>> Dipl.-Ing. Alexander Duda
>> Unterwasserrobotik
>> Robotics Innovation Center
>>
>> Hauptgeschäftsstelle Standort Bremen:
>>
>> DFKI GmbH
>> Robotics Innovation Center
>> Robert-Hooke-Straße 1
>> 28359 Bremen, Germany
>>
>> Tel.: +49 421 178 45-6620 <%2B49%20421%20178%2045-6620>
>> Zentrale: +49 421 178 45-0 <%2B49%20421%20178%2045-0>
>> Fax: +49 421 178 45-4150 <%2B49%20421%20178%2045-4150> (Faxe bitte
>> namentlich kennzeichnen)
>> E-Mail: Alexander.Duda at dfki.de
>>
>>
>> Weitere Informationen: http://www.dfki.de/robotik
>> -----------------------------------------------------------------------
>> Deutsches Forschungszentrum fuer Kuenstliche Intelligenz GmbH
>> Firmensitz: Trippstadter Straße 122, D-67663 Kaiserslautern
>> Geschaeftsfuehrung: Prof. Dr. Dr. h.c. mult. Wolfgang Wahlster
>> (Vorsitzender) Dr. Walter Olthoff
>> Vorsitzender des Aufsichtsrats: Prof. Dr. h.c. Hans A. Aukes
>> Amtsgericht Kaiserslautern, HRB 2313
>> Sitz der Gesellschaft: Kaiserslautern (HRB 2313)
>> USt-Id.Nr.: DE 148646973
>> Steuernummer: 19/673/0060/3
>>
>>
>> _______________________________________________
>> Rock-dev mailing list
>> Rock-dev at dfki.de
>> http://www.dfki.de/mailman/cgi-bin/listinfo/rock-dev
>>
>>
>
>
> --
> Vincent Vittori
> Robotic engineer MARUM
> University of Bremen
> 4 Leobener Straße, 28359 Bremen RAUM 1520
>
> *DE Mobile :* +49 152 242 37 465
> *DE Office :* +49 421 218 65 641
> *FR Mobile :* +33 612 12 35 39
> *Email:* vittori.vincent at gmail.com
> *IM:* vincent.vittori1 (Skype)
> *http://de.linkedin.com/in/vincentvittori/en
> <http://de.linkedin.com/in/vincentvittori/en>*
>
>
>
>
>
>
> --
> Dipl.-Inf. Matthias Goldhoorn
> Space and Underwater Robotic
>
> Universität Bremen
> FB 3 - Mathematik und Informatik
> AG Robotik
> Robert-Hooke-Straße 1
> 28359 Bremen, Germany
>
> Zentrale: +49 421 178 45-6611
>
> Besuchsadresse der Nebengeschäftstelle:
> Robert-Hooke-Straße 5
> 28359 Bremen, Germany
>
> Tel.: +49 421 178 45-4193
> Empfang: +49 421 178 45-6600
> Fax: +49 421 178 45-4150
> E-Mail: matthias.goldhoorn at informatik.uni-bremen.de
>
> Weitere Informationen: http://www.informatik.uni-bremen.de/robotik
>
>
> _______________________________________________
> Rock-dev mailing list
> Rock-dev at dfki.de
> http://www.dfki.de/mailman/cgi-bin/listinfo/rock-dev
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://www.dfki.de/pipermail/rock-dev/attachments/20140618/2d672403/attachment-0001.htm
More information about the Rock-dev
mailing list