>>>>> "Martin" == Martin Vermeer <[EMAIL PROTECTED]> writes:
Martin> Then you want this patch. Has the inside-inset crash fixed and
Martin> outlining buttons disabled for non-ToC. Abdel claims that
Martin> there should still be a crash in this (related to "illegal"
Martin> counters like 1.0.5).
Thanks. A few comments (some of them may be more relevant to the 1.5
version).
Index: LyXAction.C
===================================================================
--- LyXAction.C (revision 13690)
+++ LyXAction.C (working copy)
@@ -240,6 +240,7 @@ void LyXAction::init()
{ LFUN_DOWN_PARAGRAPH, "paragraph-down", ReadOnly | NoUpdate},
{ LFUN_DOWN_PARAGRAPHSEL, "paragraph-down-select", ReadOnly },
{ LFUN_GOTO_PARAGRAPH, "paragraph-goto", ReadOnly },
+ { LFUN_OUTLINE, "outline", ReadOnly },
Do you really intend this OUTLINE function to be useful for readonly
documents??
+ case LFUN_OUTLINE: {
+ lyx::toc::OutlineOp const op =
+
static_cast<lyx::toc::OutlineOp>(convert<int>(cmd.argument));
+ lyx::toc::outline(op, buffer_, cursor_.pit());
So this is a weird lfun that takes arguments 0, 1, 2, 3? What might be
better would be to have 4 different functions, that could be mapped to
kbd actions. Then I could see mergin this with LFUN_DEPTH_PLUS/MIN,
which are the same thing for environments. Note that lists would also
benefit from a way to move items (and everything nested under them) up
or down.
+ cursor_.text()->setCursor(cursor_, cursor_.pit(), 0);
+ buffer_->markDirty();
+ updateCounters(*buffer_);
+ update();
+ }
This does not look good in a case where the function did not trigger.
Of course in this case the button is disabled, but it is still a bit
ugly. Also, why do you need to reset the cursor?
outline() could return a bool telling whether something happened. Or
LFUN_OUTLINE could have proper getStatus support (which return code
would be used to enable/disable buttons).
On a separate topic, I feel some ugliness here when the cursor is not
at top-level. cursor_.pit() is not going to have the same meaning for
setCursor and outline...
Index: lfuns.h
===================================================================
--- lfuns.h (revision 13690)
+++ lfuns.h (working copy)
@@ -357,6 +357,7 @@ enum kb_action {
LFUN_BIBDB_ADD,
LFUN_BIBDB_DEL,
LFUN_INSERT_CITATION,
+ LFUN_OUTLINE, // Vermeer 20060323
LFUN_LASTACTION // end of the table
};
Spacing problem?
+void outline(OutlineOp mode, Buffer * buf, pit_type & pit)
+{
+ ParagraphList & pars = buf->text().paragraphs();
Why not pass a cursor directly? What is the information available from
the TOC itself?
+ case Up: {
...
+ case Down: {
...
I guess this is where problems can occur :)
+ case In:
+ for (; lit != lend; ++lit) {
+ if ((*lit)->toclevel == thistoclevel + 1) {
+ s->layout((*lit));
+ break;
+ }
+ }
+ break;
It could be possible/nice too to increase toclevel of all the nested
headings below (i.e. if a sections becomes subsection, have its
subsections become subsubsections). Should not be too difficult if we
ignore nested text insets.
All in all, I like this patch. I am not sure yet whether it should go
in 1.4.2, but it could go in later.
JMarc
