## Report from OpenSUSE conference 2017 ### YaST new module workshop In general better attended then the one in Dubrovnik. I think using skeleton is good, so it can more focus on explaning what is what and where to find info. Idea: Maybe it is time to revive skeleton generator for YaST modules, but more modular, as YaST already have a lot of common structures and toolings. It just should be more modular and easier to maintain. So it can e.g. generate CFA model for given file, generate new dialog, CWM widget, add travis conf and such stuff. Feedback from this part - yast_devel pattern sucks and we should update it. Also a person was hit by problem with different version of rspec gem and system ruby. Then part with YaST::Execute started. With some help everyone has working code. Feedback from this part: Silent errors are confusing for developers (when method failed, it is silently ignored and they cannot check it, they are confused what to do). As inspiration I create pull request to have in YaST::Execute also methods variants that do not show popup for errors and simply doesn't catch cheetah exception. Cheetah documentation itself missing promotion why it is better then built-in ruby shell executors. I already have it written down from one of my presentation, so it should be added there. When I talk about CFA it looks attendends are already a bit tired. I do not see much feedback from this part, probably because we already almost run out of time. Person who came here just to learn something about ruby burned out and leave after two hours. But when we talk later he said that he is impresed by ruby and plan to read something more about it. It is just too much new information to hold. Other YaST guys who were there do not hesitate to write what I forgot or what I did not notice. At the end I would like to thanks ancor for organizing it. ### My talk - YaST News Presentation was good, looks like everything is clear for audience. Having presentation after lunch break when sun is shinning really sucks. ### Interesting Parts from Other Talks Ancor's talk was well explained how our ecosystem works. For me the most intesting was question at the end of talk from audience why we do not have in Tumbleweed two yast modules that are in enterprise distribution and one that was mentioned was s390 ( so I suspect that guy was from IBM and second module is probably reipl ). As IBM is trying to make an opensuse spin for s390, maybe it make sense to also have those two modules in factory? Lada's presenantation about atom convince me to try it. Its extendability and using css for style and js for functionality really lowers barrier to adapt it for my best productivity. Just need find some time. Arvin's talk about storage-ng was good. Finally I get whole picture about design and pictures really helps to understand old and new design. Imo's talk about salt and autoyast was well presented. I think everything is clear for audience and it nicely fills picture from previous day talk about Kubic. Craig Gardner's talk about devops has interesting point, that it is good to breath a bit after delivery, which SCRUM sprint implicitelly don't do. Also it is important not just to add new cool stuff, but also drop no longer needed or efficient ones. Not just from user POV, but also in whole proces of development. libabigail talk was about automatic checking of ABI changes in ELF ( so C/C++). Interesting talk and I think it will be really interesting if it will be turn into service and write to e.g. pull requests all ABI changes, to check if it is OK. It now recognize red ( ABI incompatible), grey ( need check ) and green ( ABI compatible ) changes. Currently used in fedora. Maybe we can use it for C/C++ parts of our code? ### Feedback from venue talks YaST ncurses frontend missing mouse support even if ncurses already allows it. It is especially annoying in env which already eat some shortcuts and when there are a lot of widgets in dialog, as tab order is sometimes quite unpredictible. Feedback regarding email topic about placing lot of widgets to one dialog. In replies were multiple solutions mentioned. It would be nice to have one well supported solution for such task. Discussing with Richard Brown state of features.opensuse.org which looks for me a bit dead. We agreed that it doesn't work well and there is a plan to drop it. I provide input, that for developers who would like to help opensuse or even for our sprints planning it would be great if output of new solution is sorted list of features according to priority, so it is clear what is the most important. Problem is that openSUSE doesn't have project manager. Also history shown that voting is not solution. So how to recognize what will benefit the most of users? -- To unsubscribe, e-mail: yast-devel+unsubscribe@opensuse.org To contact the owner, e-mail: yast-devel+owner@opensuse.org