On 2009-12-10, Abdelrazak Younes wrote: > Abdelrazak Younes wrote: >> Pavel Sanda wrote: >>> Jean-Marc Lasgouttes wrote:
>>>> the real work >>>> on our buffer structure can be fast enough IMO. >>> it is not? pavel >> It is, and that is not the problem. The LateX export can happen in >> the main thread. The problem is elsewhere in that we need finer >> control of the 5 or 6 processes for latex compilation. So there's >> basically two ways of doing it: >> The easy way: let the whole latex compilation handling is done in a >> new thread -> this wouldn't even require a cloning of the Buffer. >> The hard (and maybe better) way: clean up that mess and, thanks to >> signals and slots, trigger the next latex compilation process at the >> end of the current process -> No thread needed, the different >> compilation would have to be handled in different functions callable >> independently > Just thought that I could add another and maybe best way: delegate that > 5 or 6 compilations to a python script. The only drawback would be when > latex error occurs as we would need to establish a communication > protocol between this python script and the lyx process in order to show > the errors. If you want to go this way, have a look at the "rubber" latex wrapper written in Python. It does re-running, bibtex, image conversion etc. for standalone latex compilations. http://iml.univ-mrs.fr/~beffara/soft/rubber/ Günter
