>>>>> "Lars" == Lars Gullik Bjønnes <[EMAIL PROTECTED]> writes:
Lars> Jean-Marc Lasgouttes <[EMAIL PROTECTED]> writes: >>>>>>> "Lars" == Lars Gullik Bjønnes <[EMAIL PROTECTED]> writes: >> Lars> | Lars> Feel free to find out where excatly it crashes. (and why Lars> should | Lars> it suddenly begin to crash... or did it always Lars> crash?) >> Lars> | My theory is that there is some bad memory bug somewhere and Lars> that it | gives rise to all these strang things at impredictable Lars> places. I think | the culprit is the boost::signal stuff, Lars> although the regexp report | from valgrind should be fixed. Lars> Some one should report it to boost then. Yes, but you are the one who understands so-called 'modern' C++ and boost black magic. And you are the one who decide to use this code. So somehow you cannot rely on us to clean up the mess... Of course, I could upgrade my compiler and stop to think about it. But maybe it would be more useful if you installed an older compiler and tried to play with it. Lars> | However, I won't be able to do it because (1) next week my Lars> family is | back, and I won't be able to work at home (2) I Lars> cannot compile with | debug info on my laptop (it is already Lars> very very painful) and Lars> cerr << "did I reach this place: " << var1 << var2 << endl; Lars> is a perfectly valid debug tool. Indeed. I will try to take a look. Anyway, if you look at the code in BufferView::Pimpl, you will see why I suspect the boost signal code. JMarc
