[Libreoffice-bugs] [Bug 134770] Crash deleting content since 6.3 (tracking changes enabled)

[bugzilla-daemon at bugs.documentfoundation.org]

https://bugs.documentfoundation.org/show_bug.cgi?id=134770

--- Comment #8 from Terrence Enger <lo_bugs at iseries-guru.com> ---
I see the bug in bibisect-linux-64-6.3 version oldest.  The closest
preceding version that I have is bibisect-linux-64-6.2 master, and it
procedes through the given STR without obvious problem.  Those commits
are 6 months apart, so this determination is unhelpful.


However, after bibisect-linux-64-6.2 master procedes through the given
STR, <ctrl>+Z produces Application Error (Signal 6).  Work on
debian-buster in bibisect-linux-6.1 points the introduction of this
behavior to:

          commit    s-h       date
          --------  --------  -------------------
    good  43c2ce41  b524de95  2018-04-11 14:43:30
    bad   dfd95d65  7db137e8  2018-04-11 15:11:32

and the commit message is:

    commit 7db137e87177dbe381186491ca36e3e8fd62ddc2
    Author:     Michael Stahl <Michael.Stahl at cib.de>
    AuthorDate: Wed Apr 11 14:31:47 2018 +0200
    Commit:     Michael Stahl <Michael.Stahl at cib.de>
    CommitDate: Wed Apr 11 17:11:32 2018 +0200

        tdf#50057 sw: fix duplication of at-paragraph anchored flys

        The problem is that the flys that are anchored to the first paragraph
of
        a SwRedline do get copied by SwRangeRedline::CopyToSection(), but they
        don't get deleted by DelCopyOfSection().

        Copying is enabled because the setting of the SetRedlineMove()
        was made conditional in commit 65de5382a389cc7edf1cdf506da4fb43a4d33a9f
        "#100619# fixed problem with hiding deleted graphics",
        which isn't justified in any way and which i can't imaginge a good
reason
        for, given that the flys anchored in the first paragraph are all
skipped
        in DelFlyInRange(), so why would you want to copy them; hence this
        reverts that commit.

        The interesting check for the
redline_fly_duplication_at_para_end_inside
        case is actually the one that was added in commit
        23e52c207760c596cc2f841ef59f3100c110d591

        Change-Id: I96fb294a5456e7f1172a5f408ebcb21cf211c276
        Reviewed-on: https://gerrit.libreoffice.org/52729
        Tested-by: Jenkins <ci at libreoffice.org>
        Reviewed-by: Michael Stahl <Michael.Stahl at cib.de>


As what I have bibisected is not the reported bug, I am leaving
keyword bibisectRequest.

-- 
You are receiving this mail because:
You are the assignee for the bug.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.freedesktop.org/archives/libreoffice-bugs/attachments/20200726/6a160ff4/attachment-0001.htm>