* Helmut Grohne <hel...@subdivi.de> [240706 18:51]:
> I second the change and the wording, but caution on the order and
> timing. I recommend filing all relevant problems as a MBF prior to
> changing policy. You may use dumat to gauge this problem:
> 
> SELECT * FROM content AS c1 JOIN content AS c2 JOIN package AS p1 JOIN 
> package AS p2 WHERE c1.filename LIKE 'bin/%' AND c2.filename LIKE 'sbin/%' 
> AND 's' || c1.filename = c2.filename AND c1.pid = p1.id AND c2.pid = p2.id 
> AND p1.name != p2.name;
> 
> You may download a suitable DB from
> https://subdivi.de/~helmut/dumat.sql.zst.removethis (link intentionally
> broken to prevent crawlers) or generate one yourself using dumat. You
> shall see less than 700 occurrences with significant repetition, so I
> expect fewer than 50 bug reports.

I've done additional analysis on top of this dataset, and found:

- most 'conflicts' are between packages either providing a common
  virtual package (mail-transport-agent), or otherwise provide the
  same functionality and already Conflict (hylafax/mgetty-fax,
  zfs-linux/zfs-fuse, cups/ippsample, sslh/libnet-proxy-perl).

- two packages each providing the same feature, but without
  Conflicts: today. (apache2/merecat and gearmand/gearman-server).
  These should result in two bugs.

- 10 real problems, which should result in 10 bugs:
    filename: source1/source2
    sethdlc: ax25-tools/dahdi-tools
    fai: fai/python3-flask-autoindex
    update-locale: glibc/gosa
    arcstat: nordugrid-arc-client/zfs-linux
    raddebug: freeradius/radlib
    icmd: ipmiutil/renameutils
    siggen: siggen/tripwire
    tcpd: tcp-wrappers/tcm
    tcpconnect: tcputils/bpfcc
    bitesize: ubuntu-dev-tools/bpfcc

> Once these bugs are filed and have
> their severity upgraded to at least important, I have no objections on
> including the change in policy. I do not intend to perform this work.

I'm happy to file the bugs, if policy intended to forbid this,
and/or this bug does not get rejected.

Chris

Reply via email to