Jean-Marc Lasgouttes wrote:
>>>>>> "Georg" == Georg Baum <[EMAIL PROTECTED]> writes:
>
> Georg> Not much. The stuff related to #605 is
>
> Georg> s/buffer\.temppath()/buffer.getMasterBuffer()->temppath()/g
>
> Georg> in ExternalSupport.C.
>
> But how do you deal with normal export? If file foo1.lyx includes
> dir/foo2.lyx and foo2.lyx has an external inset bar.dat, the name
> should be exported in latex as dir/bar.dat, shouldn't it?
This was indeed a missing case in my test file. I thought it should have
been bar.dat.
> From my cursory looking at the sources, we cannot do that directly in
> doSubstitution, since there are cases where we want $$FName as plain
> bar.dat.
>
> It seems to me that it is necessary to add a bool to doSubstitution
> telling wether relative file names should relative to buffer or master
> buffer.
The patch is probably wrong here for the graphics inset too. I'll have a
look.
> Georg> What I forgot to say: I did not include the "Let's pretend we
> Georg> are the master buffer" stuff because I did not understand why
> Georg> it was necessary.
>
> What is that?
I shoud really look things up before I write ;-( I meant this:
@@ -2127,9 +2127,23 @@ void Buffer::makeLaTeXFile(ostream & os,
<< endl;
texrow.newline();
}
+
+ // if we are doing a real file with body, even if this is the
+ // child of some other buffer, let's cut the link here.
+ string save_parentname;
+ if (!only_body) {
+ save_parentname = params.parentname;
+ params.parentname.erase();
+ }
+ // the real stuff
latexParagraphs(os, &*(paragraphs.begin()), 0, texrow);
+ // Restore the parenthood if needed
+ if (!only_body) {
+ params.parentname = save_parentname;
+ }
+
// add this just in case after all the paragraphs
os << endl;
texrow.newline();
> Georg> - The ExportData struct is rather a hack because it circumvents
> Georg> the const'ness of runparams. How to do this properly?
>>> Can you collect the file names at validate() time? I'd think that
>>> this pass is the right one for what you try to do.
>
> Georg> This would probably involve a call to prepareFile(), but should
> Georg> work.
>
> And would you have to call prepareFile again when generating latex?
Either that or store the result.
Georg
