Description: in tdf#104169 and tdf#142390 something happened which is very disturbing but can be easily avoided. A patch works in one operating system, but harms in another. i would suggest that besides a review by an experienced developer on his operating system at least a short check is done in all supported operating systems: 1. program can still be started with the patch, 2. program can still be operated with the patch, 3. patch does something in the direction of the described goal. Such a check could also be done by less experienced supporters. An extensive test in all systems would be too much demanded and slows down the work too much, it takes place anyway 'with the user', but gross blunders could be avoided with the above procedure. Alternatives: - the review must be done with a different operating system than the patch was developed on, or - a 'quick block option' in git, a sort of 'veto', code that is somehow harmful can be reverted or temporarily 'shut down'. Steps to Reproduce: 1. see above description Actual Results: code may harm different OS Expected Results: code should be checked first Reproducible: Always User Profile Reset: No Additional Info: targets development, thus is independent of prog ver.
Already in place, this is the role of the unit test mechanisim. Up the respective dev to add unit tests for new features. Additionally, git head master and timed release there is adequate testing of UI changes and features before they arrive in the release build. Obviously patches destined to be back porter need a harder look--and for the most part get it.