"Bo Peng" <[EMAIL PROTECTED]> writes:
| Index: src/BufferView_pimpl.C
| ===================================================================
| --- src/BufferView_pimpl.C (revision 13463)
| +++ src/BufferView_pimpl.C (working copy)
| @@ -293,6 +293,20 @@
|
| setBuffer(b);
| bv_->showErrorList(_("Parse"));
| +
| + // scroll to the position when the file was last closed
| + // FIXME: this "if(true)" should be changed to a rcfile switch
| + // I do not know how to do it now.
| + if( true ) {
| + std::pair<lyx::pit_type, lyx::pos_type> pos =
LyX::ref().lastfiles().loadFilePosition(s);
Alternatively, more similar to what you had earlier:
int pit;
int pos;
boost::tie(pit, pos) = LyX::ref().lastfiles().loadFilePosition(s);
| + // p.get<0> is pit, need to translate it to paragraph id.
pos.first
| + ParagraphList& prghs = b->paragraphs();
| + // be careful since the file may have been externally changed
...
| + if ( pos.first < prghs.size() ) {
| + saved_positions[0] = Position(s, prghs[pos.first].id(),
pos.second);
| + restorePosition(0);
| + }
| + }
Can you please drop the bookmark part of the patch for now?
If we decide that we do want bookmark[0] to be the position the cursor
was in when the doc was last saved we can put this in later.
(but... I just realized that you rely on this in your
implementation... hmm)
We actually want to get rid of the whole paragraph.id thing. It is not
stable enough. I have some (vague) plans of introducing a
(universally) unique id to use on paragraphs.
Just as a test... load a document move the cursor save.
Exit lyx. Start lyx. Load a different document. Load the first doc.
Where is the cursor placed now?
| Index: src/lastfiles.C
Doesn't this almost double the size of lastfiles.C?
(I am just trying to prove that code complexity is not reduced it IMNO
increases a lot.)
Note that despite all my negativism I am super happy that you have
picked up this ball. This is a feature I have wanted for a long time,
but have never found the time to implement.
--
Lgb
