Andre Poenitz wrote: > On Wed, Mar 17, 2004 at 03:51:14PM +0100, Georg Baum wrote: >> > Of course, if the converter command was run from the master buffer >> > dir, the situation would be different, since $$FName would always be >> > the right variable to use. >> >> This would work. I don't know very well all the capabilities of external >> templates. Angus, if this does not break anything, I propose not to >> introduce a new variable but to change the meaning of $$FName from >> >> "The filename exactly as the user entered it" >> >> to >> >> "The absolute filename (if the user entered an absolute name) or the >> filename relative to the master document (if the user entered a relative >> name)" > > I'd certainly prefer two differnt names for the two different cases. > This kind of cleverness usually backfires at some point of time.
I agree that being too clever should be avoided. However, I am not sure wether this is a case of beeing too clever. The problem is: a) A relative name in the .lyx file should be translated to a relative name in the .tex file b) An absolute name should stay absolute. a) enables the user to easily move files around (or work with several authors under cvs control), and b) is currently required in some cases if we want to output the filename without the extension. This requirement vanishes hopefully soon, but there might be others, maybe platform specific. How do you want to achieve this if you have one variable with the absolute name and one with the relative name and don't know anymore what the user entered? Of course one could define more variables (e.g. an $$AbsoluteName that is always abolute or a $$LyXName that is always exactly what the user entered, or a $$RelativeName that is always relative to the child document etc. But IMHO we need a variable with the meaning described above nevertheless. Georg
