Bug 46343

This bug is verified in KOFFICE_1_2_BRANCH.

 It does not matter what kind of text precedes the table or the tables styl=
e.=20
Nor do the columns or the font size have a bearing on this bug. It also doe=
s=20
not matter if the table is inline or not. What appears to trigger the bug i=
s=20
when the table is long enough to span two pages. I tested this in both=20
Landscape and Portrait orientation. In both cases the bug is triggered when=
=20
the table is long enough to spill over onto the next page. Here is the=20
Konsole output I captured (Koffice compiled with --enable-debug=3Dfull). In=
=20
this case the table was 36 rows long and inline:

kotext: KoStyleManager::updateGUI m_currentStyle=3D0x85aa350 New Style Temp=
late=20
(8)        //Created new style here
kotext: KoStyleManager::renameStyle 8 to New Style Template (8)
kotext: KoStyleManager::renameStyle before Standard
kotext: KoStyleManager::renameStyle after Standard
kotext: KoStyleManager::switchStyle noSignals=3Dtrue
kotext: KoStyleManager::switchStyle noSignals=3Dtrue
kotext: KoStyleManager::updateGUI updating combo to Standard
kotext: found at 0
kotext: update style Standard (0)
kword: 0x83dbde8 KWAnchor::finalize 00 paragx=3D0 paragy=3D34067
kword (tables): KWTableFrameSet::moveBy(00)
kword (formatting): KWTextFrameSet::frameResized 0x83ed638 [281724.09=20
269.5x22] invalidateLayout=3Dfalse
kword (tables): KWTableFrameSet::recalcRows (00)
kword (tables): KWTableFrameSet::recalcRows done
kword (formatting): KWTextFrameSet::frameResized 0x83ec9f8 [297.51724.09=
=20
269.5x22] invalidateLayout=3Dfalse
kword: KWAnchor::resize 11678x22151
kword: KWAnchor::resize invalidating parag 2
kword (tables): KWTableFrameSet::recalcRows (10)
kword (tables): moving 1746.09 by -2; to 1744.09
kword (tables): moving 1768.09 by -2; to 1766.09
kword (tables): moving 1790.09 by -2; to 1788.09
kword (tables): moving 1812.09 by -2; to 1810.09

<SNIP -- lines omitted>


kword (formatting): KWTextFrameSet::frameResized 0x83f24f8 [281744.09=20
269.5x22] invalidateLayout=3Dfalse
kword (tables): KWTableFrameSet::recalcRows (01)
kword (tables): KWTableFrameSet::recalcRows done
kword (formatting): KWTextFrameSet::frameResized 0x83efcb8 [297.51744.09=
=20
269.5x22] invalidateLayout=3Dfalse
kword: KWAnchor::resize 11678x25161
kword: KWAnchor::resize invalidating parag 2
kword (tables): KWTableFrameSet::recalcRows (11)
kword (tables): moving 1766.09 by -2; to 1764.09
kword (tables): moving 1788.09 by -2; to 1786.09
kword (tables): moving 1810.09 by -2; to 1808.09

<SNIP -- This pattern is repeated until I killed Kword>


I then ran Kword in gdb and killed it with ALT-CTRL-ESC and got a truly lar=
ge=20
backtrace. Here is a portion of it:

(gdb) where
#0  0x415da344 in write () from /lib/libc.so.6
#1  0x4144165c in __DTOR_END__ () from /lib/libpthread.so.0
#2  0x41395fd1 in _X11TransSocketWrite () from /usr/X11R6/lib/libX11.so.6
#3  0x41396c9d in _X11TransWrite () from /usr/X11R6/lib/libX11.so.6
#4  0x4137a09e in _XFlushInt () from /usr/X11R6/lib/libX11.so.6
#5  0x41379f9c in _XFlush () from /usr/X11R6/lib/libX11.so.6
#6  0x40673499 in XRenderChangePicture () from /usr/X11R6/lib/libXrender.so=
.1
#7  0x40b1954b in QPainter::end (this=3D0xbfffd6f0) at=20
kernel/qpainter_x11.cpp:1251
#8  0x40bad943 in QPainter::~QPainter (this=3D0xbfffd6f0 __in_chrg=3D2) at=
=20
kernel/qpainter.cpp:534
#9  0x4183a8e6 in KWCanvas::repaintChanged (this=3D0x8268be0 fs=3D0x81baf6=
0=20
resetChanged=3Dtrue) at kwcanvas.cc:172
#10 0x4186c6e7 in KWDocument::slotRepaintChanged (this=3D0x8105638=20
frameset=3D0x81baf60) at kwdoc.cc:3535
#11 0x4186b29c in KWDocument::slotRepaintVariable (this=3D0x8105638) at=20
kwdoc.cc:3350
#12 0x4186b1f6 in KWDocument::recalcVariables (this=3D0x8105638 type=3D4) =
at=20
kwdoc.cc:3317
#13 0x4186924f in KWDocument::appendPage (this=3D0x8105638) at kwdoc.cc:2803
#14 0x418be1ba in KWTableFrameSet::recalcRows (this=3D0x8316718 _col=3D0=
=20
_row=3D19) at kwtableframeset.cc:524
#15 0x418b315f in KWTextFrameSet::frameResized (this=3D0x8360dd8=20
theFrame=3D0x83657f0 invalidateLayout=3Dfalse) at kwtextframeset.cc:2171
#16 0x418b2a26 in KWTextFrameSet::slotAfterFormatting (this=3D0x8360dd8=20
bottom=3D320 lastFormatted=3D0x0 abort=3D0xbfffde5f)
    at kwtextframeset.cc:2089
#17 0x418bb487 in KWTextFrameSet::qt_invoke (this=3D0x8360dd8 _id=3D3=20
_o=3D0xbfffddd8) at kwtextframeset.moc:147
#18 0x40bab2d2 in QObject::activate_signal (this=3D0x8361d88 clist=3D0x836=
1648=20
o=3D0xbfffddd8) at kernel/qobject.cpp:2080
#19 0x41ae0720 in KoTextObject::afterFormatting (this=3D0x8361d88 t0=3D320=
=20
t1=3D0x0 t2=3D0xbfffde5f) at kotextobject.moc:208
#20 0x41add4c9 in KoTextObject::formatMore (this=3D0x8361d88 count=3D2=20
emitAfterFormatting=3Dtrue) at kotextobject.cc:1668
#21 0x41ad9b8f in KoTextObject::applyStyleChange (this=3D0x8361d88=20
changedStyle=3D0x8186888 paragLayoutChanged=3D1 formatChanged=3D0)
---Type <return> to continue or q <return> to quit---
    at kotextobject.cc:892
#22 0x418b4d9f in KWTextFrameSet::applyStyleChange (this=3D0x8360dd8=20
changedStyle=3D0x8186888 paragLayoutChanged=3D1 formatChanged=3D0)
    at kwtextframeset.cc:2511

KDE Version
1.2.94 (KDE 3.1.93 (CVS >= 20031111), compiled sources)
Operating System
FreeBSD (i386) release 4.9-RELEASE
Compiler
gcc version 2.95.4 20020320 [FreeBSD]


 This bug persists and still causes multiple pages to be generated. 

To reproduce:

1) hit enter about 5 times.
2) insert a table that has a leat 36 rows. The table will run onto the next page.
3) When this happens, kword will begin creating hundreds of additional pages.

Did as J E Drews with cvs head from 20040105. It stops at 300 pages for me. Kword behaves nice (responcive) but the rendering of the tablecells is wrong /non existant.

Cheers Jo

*** Bug 48017 has been marked as a duplicate of this bug. ***

While the multipage table feature is not wonderful or even good it does not crash anymore.

Created attachment 14995 [details]
With this file, trying to add a table kword fails

The issue is not solved, with kword 1.4.2 it occurs at least whith this file.

SVN commit 516592 by dfaure:

The infinite-loop-prevention didn't work when the paragraph was the last one of the document, due to lastFormatted==0.
CCMAIL: rafael.ontivero@gmail.com, 46343@bugs.kde.org


 M  +3 -2      KWTextFrameSet.cpp  


--- trunk/koffice/kword/KWTextFrameSet.cpp #516591:516592
@@ -2342,8 +2342,9 @@
 
         // "difference" doesn't apply if we're pasting multiple paragraphs.
         // We want to compare the height of one paragraph, not all the missing height.
-        int paragHeight = lastFormatted ? lastFormatted->rect().height() : 0;
-        kdDebug(32002) << "height we will get in the new page:" << heightWeWillGet << " parag height:" << paragHeight << endl;
+        KoTextParag* parag = lastFormatted ? lastFormatted : textDocument()->lastParag();
+        int paragHeight = parag->rect().height();
+        kdDebug(32002) << "height we will get in the new page:" << heightWeWillGet << " parag " << parag << " height:" << paragHeight << endl;
         if ( heightWeWillGet < paragHeight && !m_groupmanager )
         {
             kdDebug(32002) << "not enough height on the new page, not worth it" << endl;