Since I haven't had any response to this, I am committing it.
Cheers!
Michael D. Pritchett
On Fri, 12 Oct 2001, Michael D. Pritchett wrote:
> problem: Open a document and change the zoom. Document is now marked as
> dirty???
>
> reason: FV_View constructor calls a setCharFormat to change the language -
> which is a zero length span. This attempts to update the undo_position, but
> doesn't need to as it append to previous fmt tag. Unfortunately the
> save_position does get incremented causing a dirty document.
>
> fix: Either find a better place to do the setCharFormat or handle the case
> that if the undo_position doesn't increment neither should the
> save_position. The patch below takes the second approach.
>
> I don't think it breaks anything else, but as I don't know this area of code
> that well, I am posting to the list instead of comitting. If I hear
> nothing, I will commit by default.
>
> Cheers!
>
> Michael D. Pritchett
>
> Here is the patch:
>
> Index: pt_PT_ChangeSpan.cpp
> ===================================================================
> RCS file: /cvsroot/abi/src/text/ptbl/xp/pt_PT_ChangeSpan.cpp,v
> retrieving revision 1.32
> diff -u -r1.32 pt_PT_ChangeSpan.cpp
> --- pt_PT_ChangeSpan.cpp 2001/09/11 04:18:41 1.32
> +++ pt_PT_ChangeSpan.cpp 2001/10/12 18:32:06
> @@ -355,11 +355,13 @@
> }
> if (dpos1 == dpos2) // if length of change is zero, then we have a
> toggle format.
> {
> + UT_uint32 startUndoPos = m_history.getUndoPos();
> bool bRes = _insertFmtMarkFragWithNotify(ptc,dpos1,attributes,lProps);
> + UT_uint32 endUndoPos = m_history.getUndoPos();
> // Won't be a persistant change if it's just a toggle
> PX_ChangeRecord *pcr=0;
> m_history.getUndo(&pcr);
> - if (pcr)
> + if (pcr && (startUndoPos != endUndoPos) )
> {
> UT_DEBUGMSG(("Setting persistance of change to false\n"));
> pcr->setPersistance(false);
>
>
>
>
