Jean-Marc Lasgouttes wrote:
> Michael> Hi, with regard to bugs #575 and #438, I can provide yet
> Michael> another test case. Unfortunately, undo crashes quite often.
>
> Here is what I just wrote for #575:
>
> if (cursor.par() != tmppar
> && (cursor.par()->layout() == tmppar->layout()
> || tmppar->layout() == tclass.defaultLayoutName())
> && cursor.par()->getAlign() == tmppar->getAlign()) {
> [...]
>
> So the discussion should rather be about the usefulness of this
> feature. What would be the right behaviour in your opinion?
Hmmm, if the layout is different it might be reasonable to disallow
merging. But how about the alignment? If I delete the par break between
a centered and block aligned par, I would expect a single centered par.
What do others think?
The test case that I sent in my former email not only considers the
problem above but also points to a problem with undo (->#438). After
seeing the code above, I think bugs #575 and #438 are totally unrelated.
Juergen V. said that he cannot reproduce the bug in the test case with
1.3.0cvs. I can reproduce the bug with 1.2.2cvs whenever I want to - and
even if I don't :-)
BTW: Are there any plans to release a 1.2.2 version before 1.3.0 is out?
(I always work with the latest patches in the cvs, of course; I am just
curious)
Kind regards,
Michael
