>>>>> "Georg" == Georg Baum <[EMAIL PROTECTED]> writes:
Georg> Jean-Marc Lasgouttes wrote: >> Do you know when this can happen? Georg> For example with the attached document. >> I also thought it was not necessary. Georg> If inInset is supposed to return always a valid pointer we Georg> should assert in inInset and remove checks at other places. But Georg> then I wonder what is wrong with the current code? LyXText::read() is supposed to set it, I guess, but this never happens. Actually, since LyXText does not know the underlying inset, I am not sure how this could happen... So, when a document is read from file, setInsetOwner is not invoked correctly. It is done though for InsetTexts in read(), which calls init(). This is IMO a bug, but I am not sure of a good fix. I tried: - invoking insettext::init in Buffer::readDocument. I can't, because it is private (we could change that, though). - invoking insettextread in Buffer::readDocument. The problem is that we are supposed to return a bool telling whether we found \end_inset/\end_document. What I did not try yet: - pass the inset as parameter of LyXText::read, so that the setInsetOwner can be inoked on the paragraphs. So, what would be the best solution? JMarc PS: it is amusing to see that the various Buffer::readFile methods use a pit parameter, but it is never used... PPS: it seems that insertion of a LyX file is busted.
