Re: Crash Info

2008-01-10 Thread rgheck
Jean-Marc Lasgouttes wrote: Abdelrazak Younes <[EMAIL PROTECTED]> writes: There is a text object (Buffer::Impl::inset) but Text::pars_ is empty on construction. Maybe we should create an empty paragraph there. I see now. I you say it is safe, I believe you. But adding an empty paragra

Re: Crash Info

2008-01-10 Thread rgheck
Abdelrazak Younes wrote: rgheck wrote: It's hard to tell with all the optimization in the code, So this isn't a debug build? Yes, it is. I mean the debugger reports that the crucial variables are optimized out, so I can't be sure what they are. but it looks like the error is here: Paragr

Re: Crash Info

2008-01-10 Thread Abdelrazak Younes
Jean-Marc Lasgouttes wrote: Abdelrazak Younes <[EMAIL PROTECTED]> writes: There is a text object (Buffer::Impl::inset) but Text::pars_ is empty on construction. Maybe we should create an empty paragraph there. I see now. I you say it is safe, I believe you. But adding an empty paragraph would

Re: Crash Info

2008-01-10 Thread Jean-Marc Lasgouttes
Abdelrazak Younes <[EMAIL PROTECTED]> writes: > There is a text object (Buffer::Impl::inset) but Text::pars_ is empty > on construction. Maybe we should create an empty paragraph there. I see now. I you say it is safe, I believe you. But adding an empty paragraph would be safer (we do it on new d

Re: Crash Info

2008-01-10 Thread Abdelrazak Younes
Jean-Marc Lasgouttes wrote: Abdelrazak Younes <[EMAIL PROTECTED]> writes: Quite possibly. Text::pars_ is empty at construction and a non-loaded buffer doesn't change that fact. This commit should do it. Is this possibility due to your reorganizations? This is due to additional new checks in

Re: Crash Info

2008-01-10 Thread Jean-Marc Lasgouttes
Abdelrazak Younes <[EMAIL PROTECTED]> writes: > Quite possibly. Text::pars_ is empty at construction and a non-loaded > buffer doesn't change that fact. This commit should do it. Is this possibility due to your reorganizations? You are fixing 1 case, but I suspect many code snippets could fail in

Re: Crash Info

2008-01-10 Thread Abdelrazak Younes
rgheck wrote: It's hard to tell with all the optimization in the code, So this isn't a debug build? but it looks like the error is here: Paragraph & CursorSlice::paragraph() const { return text()->getPar(pit_); } and that it's pit_ that is invalid: pit_ is 0, but there's nothing in text

Crash Info

2008-01-10 Thread rgheck
It's hard to tell with all the optimization in the code, but it looks like the error is here: Paragraph & CursorSlice::paragraph() const { return text()->getPar(pit_); } and that it's pit_ that is invalid: pit_ is 0, but there's nothing in text(). I think rh