Hi Belén, I think that I am fine with this general design proposal.
I have some comments and questions, mostly around the details for the not-yet-written "How to set up an analysis project" content. 1) Connecting Analysis Projects to External Builds? It appears that the connection of analysis project to its build directory and artifacts is solely managed by the external build system? I assume that this is the case given the complexity and diversity of possible external build systems? If that is the case, why do we support creating such project in the Toaster GUI? Is it perhaps to support a use case where the Toaster user would get the otherwise "unattached" project's ID number (on page 5) and enter it in the external build system to complete that connection? 2) Use case for current local analysis workflow? Today you would use the tool "source toaster start" to connect a specific build directory to a Toaster build, and the "external builder" is in fact the command line user that initiates the build(s). Will we still support this general workflow? In other words, could be provide the equivalent of ... $ cd <build_dir> $ source toaster start [project_id] $ bitbake core-image-base ... where some tool will be able to (a) attach this build to <project_id> if that parameter is passed, (b) find the Toaster project for this path if it exists and add this build, or (c) create a new Toaster analysis project on the fly if this directory path is not current used? I believe that this workflow is important because it provides functional backwards compatibility, is easy to integrate for both formal and informal "build managers", and provides an easy way for reluctant command line users to get into the benefits of the deep Toaster Analysis features. All of these reasons apply to Wind River and its use of Toaster. 3) Passing data from External Builder I assume that for external Build Managers the creation and population of an Analysis Project's builds is along the same lines as you have outlined in your Jenkins summary? In other words, contrary to your statement on page 3, Toaster does not simply "collect" build information, it actually simply "receives" that information and artifacts as passed by say the plug-in for Jenkins? Specifically, it would seem that (a) during the build the analysis data would be passed directly from that remote bitbake instance to the shared Toaster database, and then (b) upon success the selected build artifacts would then be also be passed by the integration plug-in. - David > -----Original Message----- > From: [email protected] [mailto:toaster- > [email protected]] On Behalf Of Barros Pena, Belen > Sent: Thursday, May 14, 2015 8:42 AM > To: [email protected] > Subject: [Toaster] Design - dropping support for stand-alone analysis > mode (7711) > > This Bugzilla feature > > https://bugzilla.yoctoproject.org/show_bug.cgi?id=7711 > > Has considerable impact on how users interact with projects. If you > can, > please check the design document attached to the Bugzilla feature and > comment as needed. > > Thanks! > > Belén > > > -- > _______________________________________________ > toaster mailing list > [email protected] > https://lists.yoctoproject.org/listinfo/toaster -- _______________________________________________ toaster mailing list [email protected] https://lists.yoctoproject.org/listinfo/toaster
