Hi all! On Wed, Dec 11, 2024 at 09:38:41AM +0100, Helmut Grohne wrote: > I see two paths forward and we need a lasting decision on which one to > take as either of them is non-trivial effort. > > A. Give up on marking qt6-base-dev-tools as an implementation detail > and agree that the present use is ok. As a result, the package must > be split and we get practically rename qt6-base-dev-tools into > qt6-base-dev-tools-bin and then add a new qt6-base-dev-tools that > ends up being Arch:any + M-A:same + D:qt6-base-dev-tools-bin and > it contains very little beyond /usr/bin/<triplet>-qtpaths6. Also > bear in mind that since the qt6.conf resides in the qmake6 binary > package, the new qt6-base-dev-tools must also D:qmake6 enlarging it > installation set in non-trivial ways. Alternatively, qt6.conf must > be moved to yet another M-A:same package that is more fundamental. > > B. Double down on qt6-base-dev-tools and file rc bugs against 22 > source package users and 4 binary package users asking them to move > their dependency to qt6-base-dev. Then, we can add > /usr/bin/<triplet>-qtpaths6 to qt6-base-dev and stuff will work. > > Given the above, can you make a decision? Would you need any further > information?
I do not remember the reasons for recommending against depending on qt6-base-dev-tools. However, it is quite clear that people do not read the package description, and even if we fix the current 22+4 packages, we do not have a good way to prevent this from happening again with different packages. So I would go the route A. However, we should first ask Patrick, who is the primary maintainer and the author of that phrase in package description. -- Dmitry Shachnev
signature.asc
Description: PGP signature
