On Sep 9, 2014, at 9:43 AM, Brandon Allbery <[email protected]> wrote:
> > On Tue, Sep 9, 2014 at 9:27 AM, Adam Dershowitz Ph.D., P.E. > <[email protected]> wrote: > I do understand why the behavior happened, and I am not sure of the best > solution going forward. Perhaps, when doing upgrades due to “scanning > binaries for linking errors” macports should honor the full command line? I > did tell it not to upgrade openmodelica-devel, and it did it anyway. > Is there anything that I could have done differently (other then not using > -u?) > > If openmodelica-devel was rebuilt by rev-upgrade, it was because it was > affected by a link error. As such it would have been rebuilt unconditionally, > and by necessity to the latest version. I can't speak to applicability of -u > to rev-upgrade. > > In cases like this you might want to disable the automatic rev-upgrade. That > said, the fact that it was rebuilt does mean that the old version *was > broken* by some other upgrade. You may need to block upgrades of dependencies > in order to keep the older version functional. > > -- The difficulty, from a user perspective, is that it is not obvious when this kind of case will occur. I saw that I had some outdated ports, and did an upgrade of outdated and not openmodelica-devel. The difficulty is that one of those ports was omniORBpy which is what would have “broken" openmodelica-devel and caused the rev-upgrade. This is not something that would have been obvious beforehand. And also, is not a problem by itself. The problem is that “-u” got passed to this rev-upgrade so during that process, all the old ones were deleted. The idea that "sudo port -u upgrade outdated and not openmodelica-devel” will not only upgrade openmodelica-devel but also delete all the old version of that specific port seems like a trap to users. —Adam
_______________________________________________ macports-dev mailing list [email protected] https://lists.macosforge.org/mailman/listinfo/macports-dev
