On quarta-feira, 22 de junho de 2016 05:29:35 PDT Simon Hausmann wrote: > Approach 1: Roll 5.6.2 by branching 5.6 into 5.6.2. That will bring in a lot > more changes into the release and certainly requires a broad testing > effort. That is in contrary to the goal of getting a fix into the hands of > the users quickly.
Agreed, not a good idea. > Approach 2: Branch 5.6.2 off of 5.6.1, bump the version qglobal.h etc. to > 5.6.2 but just include the one fix in qtdeclarative. That is a better > approach IMO, but it still means a full rebuild of all modules. It's faster > than (1) but not as quick as (3). That would be my option. I really think we should rebuild either way. > Approach 3: Include the one declarative fix in 5.6.1, create a new tag and > rebuild declarative and all the modules that depend on it. That is the > quickest way of getting the release into the hands of the users (qtbase was > not rebuilt nor any other module not depending in declarative). We had > binaries ready for testing in under 24 hours. Note: When I wrote "rebuilt" > I mean re-compile and also re-run the auto-tests. With a big module like > qtbase you this can take a few iterations. And once qtbase changes all > depending modules undergo the same. Approach 4: change only the qtdeclarative version number (.qmake.conf) to 5.6.2 and rebuild that. All the rest stays at 5.6.1. We skip 5.6.2 on the other repositories. Drawback: this has never been tested, even though it's supposed to work. Approach 5: change qtdeclarative version number to 5.6.1.1. Same problem. > So when you say that there was no "reason" I disagree. I think there were > reasons and I was one of the people in favor of (3) that implies a -1 > suffix. I meant the use of the dash. That has nothing to do with the approach to rebuilding the affected packages. We could have used 5.6.1.1. -- Thiago Macieira - thiago.macieira (AT) intel.com Software Architect - Intel Open Source Technology Center _______________________________________________ Development mailing list Development@qt-project.org http://lists.qt-project.org/mailman/listinfo/development