I guess this whole conversation deserves a proper SIG no ? There seem to be a lot of interest and several things to talk about: standards, policies, reports, software. Actually, just came across an ABI / API breakage site for popular open source projects : http://upstream-tracker.org/ Some of the ones we use in our ecosystem are there: http://upstream-tracker.org/versions/assimp.html http://upstream-tracker.org/versions/opencv.html http://upstream-tracker.org/versions/pcl.html http://upstream-tracker.org/versions/yaml-cpp.html On Tue, Sep 24, 2013 at 6:17 PM, Dejan Pangercic wrote: > Thibault, > any company if it can it will take open source implementation, at > least as a baseline option. It can of course happen that they will > later contact a developer to get add-on features. However for now they > (Bosch is using ROS for research projects) are wary to use ROS since > there is no connection to standardized code quality indicators at all. > That of course being only one of the reasons. > > I will also say that we haven't started doing these metrics just > because we get bored here. Obviously we talk to other industrial > partners and ROScon participants and to have a way to quickly infer > software quality, possibly against some industrial standard [1] was > identified as one of the biggest needs. I know that ROS is still > predominantly research community that often couldn't care less about > such issues but that is also why we are not forcing them as the > de-facto quality view. > > > [1] - > http://portal.automotive-his.de/images/pdf/SoftwareTest/his-sc-metriken.1.3.1_e.pdf > > D. > On Tue, Sep 24, 2013 at 1:51 AM, Thibault Kruse wrote: > > > > On 24.09.2013 02:12, Dejan Pangercic wrote: > >>> > >>> Selling ROS (even for free) to a company is a marketing problem, it is > >>> about > >>> maintaining a community, a pretty website and advertising, finding > >>> reference > >>> customers speaking up for the product, and so on. > >> > >> Thibault this is unfortunately very wrong. Selling ROS is right now > >> rather a technical problem. Say you have > >> $10 available for your micro-controller. And that one has to be > >> certified, has to support some sort of encryption to protect IP, etc. > >> So you end up with the chip like this: > >> http://www.atmel.com/Images/doc6462.pdf, note 266MHz. Now you go, put > >> ROS on it and have it run drivers, task-executive, perception, > >> planning and control software. > >> Having metrics like this will not help us with getting ROS to scale to > >> something like this but it will help us convince the industry that > >> coding standards used in ROS are solid and would give us the code that > >> would run on somewhat more powerful processor(s) - which industry > >> would not have hard time to spend more money on. > >> > >> > >> Also, we do not want to force anyone to use the metrics but it is for > >> now the only way to (quickly) asses the quality of the code, > >> especially if it has been orphaned due to the lack of human resource. > > > > Hi Dejan, > > > > given that you work at Robert Bosch LLC, you kind-of represent "the > > industry", don't you (as opposed to academia)? > > And if Bosch as a company would like a vendor to provide some code, maybe > > Bosch can negotiate that directly? > > > > cheers, > > Thibault > > > > _______________________________________________ > > ros-users mailing list > > ros-users@code.ros.org > > http://lists.ros.org/mailman/listinfo/ros-users > > > > -- > Dejan Pangercic > Autonomous Technologies Group > Robert Bosch LLC > Office: +16505657403 > Cell: +16508610951 > E-Mail: dejan.pangercic@gmail.com > WWW: http://ias.cs.tum.edu/people/pangercic > _______________________________________________ > ros-users mailing list > ros-users@code.ros.org > http://lists.ros.org/mailman/listinfo/ros-users >