On 27/12/17 20:38, Jeremy Bicha wrote: > At the end of October, I uploaded an NMU for swt-gtk to drop its > unnecessary Build-Depends on webkitgtk. Only hours before the package > was set to migrate to Testing so I could ask the Release Team to > remove webkitgtk from Testing, Adrian Bunk uploaded a NMU to *add* a > depends on libwebkitgtk-1.0-0 (previously it was an undeclared > dependency). I responded by opening RC bug 880470 since swt-gtk is now > the only package keeping webkitgtk in Testing. This has now prevented > either NMU from reaching Testing. > > eclipse is the only reverse-depends of swt-gtk in Testing. I think > it's widely known that Eclipse is so ancient in Debian as to not be > very useful. However, it can't easily be removed either because there > are hundreds of packages in its rdepends chain. It also is a very big > project to update Eclipse to a current version. > > I believe that the Release Team can choose to ignore certain bugs that > would otherwise be RC. > > I request that the Release Team investigate the situation. I propose > that bug 879170 be ignored for Buster and that I be allowed to revert > the swt-gtk 3.8.2-4.2 NMU. After ~5 days, swt-gtk should migrate to > Testing, meaning that webkitgtk would no longer have any reverse > dependencies or build-dependencies in Testing and webkitgtk should be > removed from Testing then.
As much as I want to remove webkitgtk from testing, I don't think breaking existing packages makes much sense. The possible solutions I can think of after reading the bugs are: - eclipse and friends get updated to swt4-gtk - eclipse gets removed - webkit support gets disabled (this doesn't seem to be possible) Keeping a broken eclipse doesn't help, unless the breakage is small and contained. Cheers, Emilio
