Bo Peng wrote:
On Mon, May 12, 2008 at 6:43 PM, Bo Peng <[EMAIL PROTECTED]> wrote:
In that sense, the
> > reversibility problem is yours, not mine.
> >
> I'm perplexed about this, since I don't have any such problem.
Let me try it the last time.
Let me try to summarize the options:
Bo's embedding proposal:
1) This is about embedding files directly in the '.lyx' file encoded
with base64.
2) Each and every external file can be embedded individually.
3) By default, the original filename is not put in the '.lyx'.
4) Optionally and strictly per user request and per inset, the original
filename can be put encrypted in the '.lyx' file.
5) Each and every embedded object can be extracted to an external file:
5-a) There is no \filename tag: the object is extracted in the same
directory as the '.lyx' file (or maybe in '.lyxdir/'?) and with a
default file name (ex: object1.png).
5-b) There is a \filename tag: the user is asked for comfirmation and is
asked for a password.
Richard's bundling proposal:
1) This is about bundling together all necessary files for proper
compilation of the '.lyx' file.
2) There is only one \bundle tag for the '.lyx' file. Insets are not
bundled/unbundled individually.
3) When bundling, out of tree files are copied in 'lyxdir/'. Original
file directory imformation is lost, only file names are preserved.
4) Per inset, Bundled files can be updated with out of tree files.
4) Original file directory information can be optionally put in the
session info.
6) If 5) is saved, the original '.lyx' file can be restored and the
external file copied back to their original directory. Maybe the user
should be offered a list of external files together with check boxes.
Do I get the proposals right? If yes, I think the two proposals are
perpendicular WRT implementation and they could both be offered to the
user. They respond to two different use cases IMHO.
Abdel.
