Mattias Gaertner wrote 6 lines
> I'm still debugging, but I get the feeling this is because someone passes
> some fpmake as an unit (to document) while it is a pascal main program. Of
> course fpdoc could check better, but probably all it can do is throw an
> error.
That would be much better than a crash.
For now, I only added a check there, since I don't where to thoroughly fix this. (it should probably happen much earlier, probably at the "program" kw)
> So probably the scripts that auto create lists of units should filter out
> fpmake.
yes
Did that too, build_lcl_docs.lpr
I still have some exceptions at the end of the compile sometimes. Luckily I have been able to find a root cause of them (unitialized variables, -gttt and -CR have been very helpful, as usual).
Most problems seem to be in TClassTreeBuilder.AddToClassTree.
Also the CHMs are not as nicely hyperlinked as the 2.6.0 builds were, somehow the .xct content isn't resolved always in again the classtree stuff.