Angus wrote: > On Tuesday 05 August 2003 8:38 am, Alfredo Braunstein wrote: > > > 2. Rewrite the logic as a shell script and just use the forked call > > > stuff. > > > > This is an option. But the LyX LaTeX code seems to be full of tricks and > > hacks I don't understand, so it's completely out of reach for me. > > I will definitely agree with you there. It's horrible. > > > > This is for the future though, don't you think? ;-) > > > > I agree, but knowing destination can help to find the path, don't you > > think? ;-) > > Agree again, but I think that using threads/forked processes is an > implementation detail that we should hide out of sight. I think that the > first step is to clean up the code in the insets so that Inset::latex does > one thing and one thing only: write a snippet of latex to the ostream. > > In order to achieve that we need a new member function to perform any > necessary manipulation of the external material. To my mind, it makes sense > here to write these instructions to an ostream also. We end up with two > separate ostreams, one containing instructions on how to manipulate external > files and one containing the latex/docbook document. > > To get to this point will require _no_ interaction with the outside world and > places us in a strong position to contemplate how best to do so. > > Don't you think?
As always, you are right. It would/will be a nice cleanup in any case. :-) Regards, Alfredo