<html>
  <head>
    <meta content="text/html; charset=ISO-8859-1"
      http-equiv="Content-Type">
  </head>
  <body bgcolor="#FFFFFF" text="#000000">
    Hi Michael, Cedric,<span id="result_box" class="" lang="en"><span><br>
      </span></span><br>
    <meta http-equiv="content-type" content="text/html;
      charset=ISO-8859-1">
    Thank you for everything.<br>
    <br>
    (11/11/19 7:17), Michael Stahl wrote:<br>
    <blockquote cite="mid:ja6li3$2s2$1@dough.gmane.org" type="cite"><br>
      <pre wrap="">hi Tomofumi,

thank you for your patch; i have pushed it, but not as a fix to the bug
40831  :)
</pre>
    </blockquote>
    (--)<br>
    <blockquote cite="mid:ja6li3$2s2$1@dough.gmane.org" type="cite">
      <blockquote type="cite">
        <blockquote type="cite">
          <blockquote type="cite">
            <pre wrap="">i am currently not sure if your patch is the right fix; i would assume
that the cursor that points to the removed node should have been moved
away from it by ... something.
</pre>
          </blockquote>
        </blockquote>
      </blockquote>
      <pre wrap="">really the problem here is that we cannot store the position as a bunch
of integers or a non-relocatable SwPosition here where nodes may
actually be deleted; we need a cursor that is automatically corrected by
PamCorrAbs/Rel.

this is fixed by using SwUnoCrsr instead:
<a class="moz-txt-link-freetext" href="http://cgit.freedesktop.org/libreoffice/core/commit/?id=82a716f29cc252740c80556f72f9d9e602877918">http://cgit.freedesktop.org/libreoffice/core/commit/?id=82a716f29cc252740c80556f72f9d9e602877918</a>
</pre>
    </blockquote>
    <br>
    <span id="result_box" class="" lang="en"><span class="hps">I think</span>
      <span class="hps">much better</span> <span class="hps">of</span>
      this fix<span class="hps"></span></span><span id="result_box"
      class="" lang="en"><span class="hps"></span><span> than mine,too.</span></span><br>
    <br>
    <br>
    <blockquote cite="mid:ja6li3$2s2$1@dough.gmane.org" type="cite">
      <blockquote type="cite">
        <blockquote type="cite">
          <pre wrap="">Well, Tomofumi's patch looks like on more safety check in that couldn't
harm. I'ld go for pushing his (second) patch.
</pre>
        </blockquote>
      </blockquote>
      <pre wrap="">indeed, our users would probably appreciate it if writer does not crash
when this kind of problem occurs, so i have pushed Tomofumi's patch to
prevent this from happening:
<a class="moz-txt-link-freetext" href="http://cgit.freedesktop.org/libreoffice/core/commit/?id=af0333e5865ecd184270af3211880d6d02307c31">http://cgit.freedesktop.org/libreoffice/core/commit/?id=af0333e5865ecd184270af3211880d6d02307c31</a>

but then i added assertions to detect the problem, so that this
work-around will not cover it up completely and hopefully developers pay
attention:
<a class="moz-txt-link-freetext" href="http://cgit.freedesktop.org/libreoffice/core/commit/?id=9de2110e5bed17fcc6050bfaa669df45f0c1eb44">http://cgit.freedesktop.org/libreoffice/core/commit/?id=9de2110e5bed17fcc6050bfaa669df45f0c1eb44</a>
</pre>
    </blockquote>
    <br>
    I also agree with.<br>
    Thanks again.<br>
    <br>
    <br>
    Best regards,<br>
    --<br>
    Tomofumi Yagi<br>
    <br>
  </body>
</html>