Jean-Marc,
I have been mulling over how best to overcome the limitations in my new
FileName class that you pointed out in Chemnitz. I think that this
algorithm should work well:
The FileName class should be able to handle three possibilities
1. Input/Output the absolute filename.
Store
string file = <absolute filename>;
bool output_absolute_filename = true;
2. Input/Output the filename relative to Buffer::filePath().
Store
string file = <absolute filename>;
bool output_absolute_filename = false;
Pass the output function Buffer::filePath() or set Path appropriately.
3. Input/Output a relative filename. The file is found using TeX magic
(TEXINPUTS, BIBINPUTS etc).
string file = <relative filename>;
bool output_absolute_filename = false;
When inputing a new bibtex database in the bibtex dialog, we check whether a
filename with relative path can be found with the Buffer::filePath. If it
can, we use case 2. If not, we test whether TeX can find it. If it can, we
use case 3, else we use case 2 and assume that the file does not exist.
In all three cases, if the inset is cut and pasted to another buffer, then
its contents do not change.
--
Angus
