Sergio Gelato <sergio.gel...@astro.su.se> writes: > That's what my suggested Breaks: addition was intended to document and > enforce. > > IIRC the issue came to my attention while backporting openafs 1.8.0~pre1 > to jessie from experimental. One then has a choice between reverting a > patch to openafs (which will then break when libroken is upgraded) or > upgrading libroken as well (which will break old clients).
I asked about this on the #debian-release IRC channel, as they would have to approve any solution for inclusion in Stretch. They feel this is a release critical bug. However they don't like your solution of using the breaks header. I have submitted a bug report upstream: https://github.com/heimdal/heimdal/issues/279 I am inclined to think the best solution for now will be to patch in aliases for the two symbols that were removed. Also reasons for renaming the symbol in the first place are documented here: https://github.com/heimdal/heimdal/issues/107 In this case I don't think interference with gcrypt is likely, as the only package in Debian that uses Heimdal is OpenAFS, and neither Heimdal or OpenAFS depend on heimdal. Might be a problem - at least in theory - for locally built packages however. -- Brian May <b...@debian.org>
